occupancy_learning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "#import seaborn as sns\n",
    "import numpy as np\n",
    "import uuid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from rdflib import ConjunctiveGraph, Graph, Namespace, URIRef, Literal\n",
    "import rdflib\n",
    "XSD = Namespace('http://www.w3.org/2001/XMLSchema#')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.3.0\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.layers.experimental import preprocessing\n",
    "\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow_docs as tfdocs\n",
    "import tensorflow_docs.plots\n",
    "import tensorflow_docs.modeling\n",
    "\n",
    "import os\n",
    "os.environ['KMP_DUPLICATE_LIB_OK']='True'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_and_compile_model(norm):\n",
    "    model = keras.Sequential([\n",
    "        norm,\n",
    "        layers.Dense(64, activation='relu'),\n",
    "        layers.Dense(64, activation='relu'),\n",
    "        layers.Dense(1)\n",
    "    ])\n",
    "\n",
    "    model.compile(\n",
    "        loss='mse',\n",
    "        optimizer=tf.keras.optimizers.Adam(0.001),\n",
    "        # use complete metric names\n",
    "        metrics=['mean_absolute_error', 'mean_squared_error'])\n",
    "    return model\n",
    "\n",
    "\n",
    "def build_model(input_size):\n",
    "  model = keras.Sequential([\n",
    "    layers.Dense(64, activation='relu', input_shape=[input_size]),\n",
    "    layers.Dense(64, activation='relu'),\n",
    "    layers.Dense(1)\n",
    "  ])\n",
    "\n",
    "  optimizer = tf.keras.optimizers.RMSprop(0.001)\n",
    "\n",
    "  model.compile(loss='mse',\n",
    "                optimizer=optimizer,\n",
    "                metrics=['mae', 'mse']) # 'acc', tf.keras.metrics.MeanSquaredError()\n",
    "\n",
    "  return model\n",
    "\n",
    "\"\"\" displays prediction diagram by applying model of source room to \n",
    "data of target room\n",
    "\"\"\"\n",
    "def prediction_graph(source_room, target_room):\n",
    "    \n",
    "    test_predictions = rooms[source_room]['model'].predict(rooms[target_room]['normed_test_data']).flatten()\n",
    "\n",
    "    a = plt.axes(aspect='equal')\n",
    "    plt.scatter(rooms[target_room]['test_labels'], test_predictions)\n",
    "    plt.xlabel('True Values [Occupants]')\n",
    "    plt.ylabel('Predictions [Occupants]')\n",
    "    lims = [0, 80]\n",
    "    plt.xlim(lims)\n",
    "    plt.ylim(lims)\n",
    "    _ = plt.plot(lims, lims)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\" Merge time columns based on given interval\n",
    "\"\"\"\n",
    "def merge_time(input_df, data_column, interval = 30, time_filter = True):\n",
    "    input_df['merged_time'] = input_df.apply(lambda row: str(row['year']) +\n",
    "                         str(row['month']).zfill(2) + \n",
    "                         str(row['day']).zfill(2) +\n",
    "                         row['time'][:2] + \n",
    "                         str(int(row['time'][3:5]) // interval).zfill(2), axis=1)\n",
    "    \n",
    "    if time_filter:\n",
    "        drop_index = input_df[input_df['time'].map(lambda row: int(row[:2]) < 6 or int(row[:2]) > 14)].index\n",
    "        input_df.drop(drop_index , inplace=True)\n",
    "    \n",
    "    \n",
    "    return input_df.groupby(['merged_time'])[data_column].mean()\n",
    "    \n",
    "\n",
    "\"\"\" normalize data\n",
    "\"\"\"\n",
    "def norm_xxx(x, x_stats):\n",
    "  return (x - x_stats['mean']) / x_stats['std']\n",
    "    \n",
    "    \n",
    "\"\"\" build dataset from raw data\n",
    "\"\"\"\n",
    "def build_dataset(room):\n",
    "\n",
    "    ### filled\n",
    "    occupancy_file = \"/Users/amin/Downloads/Dataset/filleddata/occupant_count_room_%d.csv\"%room\n",
    "    co2_file = \"/Users/amin/Downloads/Dataset/filleddata/co2_room_%d.csv\"%room\n",
    "    temperature_file = \"/Users/amin/Downloads/Dataset/filleddata/temperature_room_%d.csv\"%room\n",
    "    #illuminance_file = \"/Users/amin/Downloads/Dataset/filleddata/Illuminance_room_%d.csv\"%room\n",
    "    humidity_file = \"/Users/amin/Downloads/Dataset/filleddata/humidity_room_%d.csv\"%room\n",
    "\n",
    "    ### original\n",
    "    #occupancy_file = \"/Users/amin/Downloads/Dataset/original/occupant_count_room_1.csv\"\n",
    "    #co2_file = \"/Users/amin/Downloads/Dataset/original/co2_room_1.csv\"\n",
    "\n",
    "    occ_columns = ['year', 'month', 'time', 'day', 'workday', 'occupancy']\n",
    "    co2_columns = ['year', 'month', 'time', 'day', 'workday', 'co2']\n",
    "    #tmp_columns = ['year', 'month', 'time', 'day', 'workday', 'temperature']\n",
    "    #ill_columns = ['year', 'month', 'time', 'day', 'workday', 'illuminance']\n",
    "    #hum_columns = ['year', 'month', 'time', 'day', 'workday', 'humidity']\n",
    "\n",
    "    occ_df = pd.read_csv(occupancy_file, names=occ_columns, skiprows=1)\n",
    "    co2_df = pd.read_csv(co2_file, names=co2_columns, skiprows=1)\n",
    "    #tmp_df = pd.read_csv(temperature_file, names=tmp_columns, skiprows=1)\n",
    "    #ill_df = pd.read_csv(illuminance_file, names=ill_columns, skiprows=1)\n",
    "    #hum_df = pd.read_csv(humidity_file, names=hum_columns, skiprows=1)\n",
    "    \n",
    "    \n",
    "    occ_grouped = merge_time(occ_df, 'occupancy')\n",
    "    co2_grouped = merge_time(co2_df, 'co2')\n",
    "    #tmp_grouped = merge_time(tmp_df, 'temperature')\n",
    "    #ill_grouped = merge_time(ill_df, 'illuminance')\n",
    "    #hum_grouped = merge_time(hum_df, 'humidity')\n",
    "    \n",
    "    \n",
    "    #dataset = pd.concat([occ_grouped, co2_grouped, tmp_grouped, ill_grouped, hum_grouped], axis=1, sort=False)\n",
    "    dataset = pd.concat([occ_grouped, co2_grouped], axis=1, sort=False)\n",
    "    dataset['occupancy'] = dataset['occupancy'].round()\n",
    "\n",
    "    # simplifying occupancy in 10X groups \n",
    "    #dataset['occupancy'] = dataset['occupancy']//10\n",
    "\n",
    "\n",
    "    # simplifying occupancy in 20X groups \n",
    "    #dataset['occupancy'] = dataset['occupancy']//20\n",
    "\n",
    "    # simplifying occupancy in 5X groups \n",
    "    #dataset['occupancy'] = dataset['occupancy']//5 \n",
    "    \n",
    "    return dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "     count        mean         std         min        25%         50%  \\\n",
      "co2  634.0  646.489927  133.201777  450.613337  521.43466  634.767994   \n",
      "\n",
      "           75%         max  \n",
      "co2  757.16534  992.810677  \n"
     ]
    }
   ],
   "source": [
    "room_id = 1\n",
    "dataset = room_dataset[room_id].copy()\n",
    "train_dataset = dataset.sample(frac=0.8,random_state=0)\n",
    "test_dataset = dataset.drop(train_dataset.index)\n",
    "\n",
    "# dataset statistics\n",
    "train_stats = train_dataset.describe()\n",
    "train_stats.pop(\"occupancy\")\n",
    "train_stats = train_stats.transpose()\n",
    "print(train_stats)\n",
    "\n",
    "#split featured from lables\n",
    "train_labels = train_dataset.pop('occupancy')\n",
    "test_labels = test_dataset.pop('occupancy')\n",
    "\n",
    "# normalize the data\n",
    "normed_train_data = norm(train_dataset, train_stats)\n",
    "normed_test_data = norm(test_dataset, train_stats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 64)                128       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                4160      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 4,353\n",
      "Trainable params: 4,353\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "\n",
      "Epoch: 0, acc:0.1203,  loss:807.5320,  mae:20.1766,  mse:807.5320,  val_acc:0.0315,  val_loss:806.8320,  val_mae:20.1976,  val_mse:806.8320,  \n",
      ".........................................................................."
     ]
    }
   ],
   "source": [
    "model = build_model(len(train_dataset.keys()))\n",
    "model.summary()\n",
    "\n",
    "EPOCHS = 1000\n",
    "\n",
    "# The patience parameter is the amount of epochs to check for improvement\n",
    "early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)\n",
    "\n",
    "history = model.fit(normed_train_data, train_labels, \n",
    "                    epochs=EPOCHS, validation_split = 0.2, verbose=0, \n",
    "                    callbacks=[early_stop, tfdocs.modeling.EpochDots()])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>loss</th>\n",
       "      <th>mae</th>\n",
       "      <th>mse</th>\n",
       "      <th>acc</th>\n",
       "      <th>val_loss</th>\n",
       "      <th>val_mae</th>\n",
       "      <th>val_mse</th>\n",
       "      <th>val_acc</th>\n",
       "      <th>epoch</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>212.747009</td>\n",
       "      <td>10.358898</td>\n",
       "      <td>212.747009</td>\n",
       "      <td>0.100592</td>\n",
       "      <td>243.602295</td>\n",
       "      <td>11.286337</td>\n",
       "      <td>243.602295</td>\n",
       "      <td>0.031496</td>\n",
       "      <td>69</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>212.699753</td>\n",
       "      <td>10.388831</td>\n",
       "      <td>212.699753</td>\n",
       "      <td>0.100592</td>\n",
       "      <td>243.540222</td>\n",
       "      <td>11.277650</td>\n",
       "      <td>243.540222</td>\n",
       "      <td>0.031496</td>\n",
       "      <td>70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>212.957581</td>\n",
       "      <td>10.399727</td>\n",
       "      <td>212.957581</td>\n",
       "      <td>0.100592</td>\n",
       "      <td>243.541702</td>\n",
       "      <td>11.306745</td>\n",
       "      <td>243.541702</td>\n",
       "      <td>0.031496</td>\n",
       "      <td>71</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>212.316315</td>\n",
       "      <td>10.388264</td>\n",
       "      <td>212.316315</td>\n",
       "      <td>0.100592</td>\n",
       "      <td>243.758057</td>\n",
       "      <td>11.234996</td>\n",
       "      <td>243.758057</td>\n",
       "      <td>0.031496</td>\n",
       "      <td>72</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>212.341553</td>\n",
       "      <td>10.381507</td>\n",
       "      <td>212.341553</td>\n",
       "      <td>0.100592</td>\n",
       "      <td>243.811005</td>\n",
       "      <td>11.228868</td>\n",
       "      <td>243.811005</td>\n",
       "      <td>0.031496</td>\n",
       "      <td>73</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          loss        mae         mse       acc    val_loss    val_mae  \\\n",
       "69  212.747009  10.358898  212.747009  0.100592  243.602295  11.286337   \n",
       "70  212.699753  10.388831  212.699753  0.100592  243.540222  11.277650   \n",
       "71  212.957581  10.399727  212.957581  0.100592  243.541702  11.306745   \n",
       "72  212.316315  10.388264  212.316315  0.100592  243.758057  11.234996   \n",
       "73  212.341553  10.381507  212.341553  0.100592  243.811005  11.228868   \n",
       "\n",
       "       val_mse   val_acc  epoch  \n",
       "69  243.602295  0.031496     69  \n",
       "70  243.540222  0.031496     70  \n",
       "71  243.541702  0.031496     71  \n",
       "72  243.758057  0.031496     72  \n",
       "73  243.811005  0.031496     73  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hist = pd.DataFrame(history.history)\n",
    "hist['epoch'] = history.epoch\n",
    "hist.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build datasets and models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [],
   "source": [
    "# build datasets\n",
    "room_dataset = {1: build_dataset(1), 2:build_dataset(2), 3:build_dataset(3)}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch: 0, loss:846.9637,  mean_absolute_error:21.6645,  mean_squared_error:846.9637,  val_loss:583.5036,  val_mean_absolute_error:14.3402,  val_mean_squared_error:583.5036,  \n",
      ".................Room_ 1 (ds_id:  b782a3a359e54fcebbd57d590665bc39 , model_id: 0754017c705548d490065f71ac4364ee )\n",
      "\n",
      "Epoch: 0, loss:160.7701,  mean_absolute_error:9.5912,  mean_squared_error:160.7701,  val_loss:81.8918,  val_mean_absolute_error:7.0348,  val_mean_squared_error:81.8918,  \n",
      "..................................Room_ 2 (ds_id:  49117df88973460f947218e5f0adb18e , model_id: 24e6a2b6b9c64aec9605bc00b2ed3b5a )\n",
      "\n",
      "Epoch: 0, loss:183.0975,  mean_absolute_error:9.5052,  mean_squared_error:183.0975,  val_loss:99.0436,  val_mean_absolute_error:7.6095,  val_mean_squared_error:99.0436,  \n",
      "................Room_ 3 (ds_id:  fa639abdd2e940149afd422aab51e5d8 , model_id: cac90a7b1bf147bda94625edc5b339f7 )\n"
     ]
    }
   ],
   "source": [
    "# build models\n",
    "rooms = {}\n",
    "\n",
    "for room_id in [1,2,3]:\n",
    "    \n",
    "    # build dataset\n",
    "    dataset = room_dataset[room_id].copy()\n",
    "    \n",
    "    #train_dataset = dataset.sample(frac=0.8,random_state=0)\n",
    "    #test_dataset = dataset.drop(train_dataset.index)\n",
    "\n",
    "    test_dataset = dataset.sample(n=100,random_state=0)\n",
    "    train_dataset = dataset.drop(test_dataset.index)\n",
    "    finetune_dataset = train_dataset.sample(n=100,random_state=0)\n",
    "    \n",
    "    #retrain_dataset = dataset.sample(n=100,random_state=0)\n",
    "    #retrain_labels = retrain_dataset.pop('occupancy')\n",
    "    #rest_dataset = dataset.drop(retrain_dataset.index)\n",
    "    #test_dataset = rest_dataset.sample(n=100,random_state=0)\n",
    "    \n",
    "    \n",
    "    \n",
    "    ds_uuid = uuid.uuid4().hex\n",
    "    ds_file = 'dataset/ds_{}.h5'.format(ds_uuid)\n",
    "    store = pd.HDFStore(ds_file)\n",
    "    store['train'] = train_dataset\n",
    "    store['finetune'] = finetune_dataset\n",
    "    store['test'] = test_dataset\n",
    "    store.close()\n",
    "    \n",
    "\n",
    "    #split featured from lables\n",
    "    train_labels = train_dataset.pop('occupancy')\n",
    "    test_labels = test_dataset.pop('occupancy')\n",
    "\n",
    "\n",
    "    # normalization layer\n",
    "    co2_feature = np.array(train_dataset['co2'])\n",
    "    co2_normalizer = preprocessing.Normalization(input_shape=[1,])\n",
    "    co2_normalizer.adapt(co2_feature)\n",
    "    \n",
    "    model = build_and_compile_model(co2_normalizer)\n",
    "    \n",
    "    \n",
    "    #model.summary()\n",
    "\n",
    "    EPOCHS = 1000\n",
    "\n",
    "    # The patience parameter is the amount of epochs to check for improvement\n",
    "    early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)\n",
    "\n",
    "    history = model.fit(train_dataset, train_labels, \n",
    "                        epochs=EPOCHS, validation_split = 0.2, verbose=0, \n",
    "                        callbacks=[early_stop, tfdocs.modeling.EpochDots()])\n",
    "    \n",
    "    # store model\n",
    "    model_uuid = uuid.uuid4().hex\n",
    "    \n",
    "    print('Room_',room_id,'(ds_id: ', ds_uuid, ', model_id:', model_uuid, ')')\n",
    "    \n",
    "    model_file = 'model/model_{}.h5'.format(model_uuid)\n",
    "    model.save(model_file)\n",
    "\n",
    "    #loss, mae, mse, acc = model.evaluate(normed_test_data, test_labels, verbose=2)\n",
    "    rooms[room_id] = {\n",
    "        'test_data' : test_dataset, \n",
    "        'test_labels' : test_labels, \n",
    "        'model': model, \n",
    "        'model_file' : model_file, \n",
    "        'dataset_uuid': ds_uuid,\n",
    "        'model_uuid': model_uuid\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 575,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'MAE [Occupants]')"
      ]
     },
     "execution_count": 575,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plotter = tfdocs.plots.HistoryPlotter(smoothing_std=2)\n",
    "plotter.plot({'Basic': history}, metric = 'mean_absolute_error')\n",
    "plt.ylim([0, 20])\n",
    "plt.ylabel('MAE [Occupants]')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 40.5253 - mean_absolute_error: 5.0246 - mean_squared_error: 40.5253\n",
      "Testing set Mean Abs Error:  5.02 CO2\n"
     ]
    }
   ],
   "source": [
    "room_id = 3\n",
    "loss, mae, mse = rooms[room_id]['model'].evaluate(rooms[room_id]['test_data'], rooms[room_id]['test_labels'], verbose=2)\n",
    "print(\"Testing set Mean Abs Error: {:5.2f} CO2\".format(mae))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 585,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_predictions = model.predict(rooms[room_id]['test_data']).flatten()\n",
    "\n",
    "a = plt.axes(aspect='equal')\n",
    "plt.scatter(test_labels, test_predictions)\n",
    "plt.xlabel('True Values [Occupants]')\n",
    "plt.ylabel('Predictions [Occupants]')\n",
    "lims = [0, 80]\n",
    "plt.xlim(lims)\n",
    "plt.ylim(lims)\n",
    "_ = plt.plot(lims, lims)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 586,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#error = test_predictions - test_labels\n",
    "error = test_predictions - rooms[room_id]['test_labels']\n",
    "plt.hist(error, bins = 25)\n",
    "plt.xlabel(\"Prediction Error [Occupants]\")\n",
    "_ = plt.ylabel(\"Count\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "occupancy      0\n",
       "co2            0\n",
       "temperature    0\n",
       "humidity       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Preparing dataset for overlay SNS diagram\n",
    "room1_dataset = room_dataset[1].copy()\n",
    "room2_dataset = room_dataset[2].copy()\n",
    "room3_dataset = room_dataset[3].copy()\n",
    "\n",
    "room1_dataset['room'] = 1\n",
    "room2_dataset['room'] = 2\n",
    "room3_dataset['room'] = 3\n",
    "\n",
    "dataset=room1_dataset.copy()\n",
    "dataset = dataset.append(room2_dataset)\n",
    "dataset = dataset.append(room3_dataset)\n",
    "\n",
    "sns.pairplot(dataset, diag_kind=\"kde\", hue=\"room\",  plot_kws=dict(alpha=0.2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Drawing prediction subplots\n",
    "fig = plt.figure()\n",
    "fig, axs = plt.subplots(3, 3, sharex=True, sharey=True, figsize=(10, 10))\n",
    "\n",
    "lims = [0, 80]\n",
    "plt.xlim(lims)\n",
    "plt.ylim(lims)\n",
    "\n",
    "cols = ['Room 1 data', 'Room 2 data', 'Room 3 data']\n",
    "rows = ['Room 1 model', 'Room 2 model', 'Room 3 model']\n",
    "\n",
    "for ax, col in zip(axs[0], cols):\n",
    "    ax.set_title(col)\n",
    "\n",
    "for ax, row in zip(axs[:,0], rows):\n",
    "    ax.set_ylabel(row, size='large')\n",
    "\n",
    "fig.tight_layout()\n",
    "\n",
    "\n",
    "for source_room in [1,2,3]:\n",
    "    for target_room in [1,2,3]:\n",
    "        test_predictions = rooms[source_room]['model'].predict(rooms[target_room]['test_data']).flatten()\n",
    "        axs[source_room - 1 , target_room - 1].scatter(rooms[target_room]['test_labels'], test_predictions)\n",
    "        _ = axs[source_room - 1 , target_room - 1].plot(lims, lims)\n",
    "        #axs[source_room - 1, target_room -1].set_xlabel('True Values [Occupants]')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 167.9990 - mean_absolute_error: 9.7587 - mean_squared_error: 167.9990\n",
      "4/4 - 0s - loss: 63.6485 - mean_absolute_error: 5.4978 - mean_squared_error: 63.6485\n",
      "4/4 - 0s - loss: 129.4320 - mean_absolute_error: 9.6490 - mean_squared_error: 129.4320\n",
      "4/4 - 0s - loss: 327.7086 - mean_absolute_error: 13.4043 - mean_squared_error: 327.7086\n",
      "4/4 - 0s - loss: 21.8347 - mean_absolute_error: 3.7022 - mean_squared_error: 21.8347\n",
      "4/4 - 0s - loss: 65.6372 - mean_absolute_error: 6.6769 - mean_squared_error: 65.6372\n",
      "4/4 - 0s - loss: 328.2416 - mean_absolute_error: 13.3242 - mean_squared_error: 328.2416\n",
      "4/4 - 0s - loss: 34.1794 - mean_absolute_error: 4.6656 - mean_squared_error: 34.1794\n",
      "4/4 - 0s - loss: 40.5407 - mean_absolute_error: 5.0176 - mean_squared_error: 40.5407\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Drawing error subplots\n",
    "fig = plt.figure()\n",
    "fig, axs = plt.subplots(3, 3, sharex=True, sharey=True, figsize=(10, 10))\n",
    "\n",
    "cols = ['Room 1 data', 'Room 2 data', 'Room 3 data']\n",
    "rows = ['Room 1 model', 'Room 2 model', 'Room 3 model']\n",
    "\n",
    "for ax, col in zip(axs[0], cols):\n",
    "    ax.set_title(col)\n",
    "\n",
    "for ax, row in zip(axs[:,0], rows):\n",
    "    ax.set_ylabel(row, size='large')\n",
    "\n",
    "fig.tight_layout()\n",
    "\n",
    "cnt = 1\n",
    "for source_room in [1,2,3]:\n",
    "    \n",
    "    source_model = rooms[source_room]['model']\n",
    "\n",
    "    for target_room in [1,2,3]:\n",
    "        \n",
    "        loss, mae, mse = source_model.evaluate(rooms[target_room]['test_data'], rooms[target_room]['test_labels'], verbose=2)\n",
    "        axs[source_room - 1 , target_room - 1].text(15, 20, ' mae: {:5.2f} \\n mse: {:5.2f}'.format(mae, mse), style='italic',\n",
    "            bbox={'facecolor': 'black', 'alpha': 0.05, 'pad': 5})\n",
    "        \n",
    "        test_predictions = source_model.predict(rooms[target_room]['test_data']).flatten()\n",
    "        error = test_predictions - rooms[target_room]['test_labels'] \n",
    "        axs[source_room - 1 , target_room - 1].hist(error, bins = 25)\n",
    "        \n",
    "        #plt.xlabel(\"Prediction Error [Occupants]\")\n",
    "        #_ = plt.ylabel(\"Count\")\n",
    "        cnt += 1\n",
    "    \n",
    "        \n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'loss': <tf.Tensor: shape=(), dtype=float32, numpy=48.950825>,\n",
       " 'mae': <tf.Tensor: shape=(), dtype=float32, numpy=5.2487955>,\n",
       " 'mse': <tf.Tensor: shape=(), dtype=float32, numpy=48.950825>,\n",
       " 'acc': <tf.Tensor: shape=(), dtype=float32, numpy=0.10236221>}"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "{m.name: m.result() for m in tf_model.metrics}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_6 (Dense)              (None, 64)                128       \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 64)                4160      \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 4,353\n",
      "Trainable params: 4,353\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "tf_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Add Model to Ontology"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "#RDF        = Namespace('http://www.w3.org/1999/02/22-rdf-syntax-ns#')\n",
    "#RDFS       = Namespace('http://www.w3.org/2000/01/rdf-schema#')\n",
    "#OWL        = Namespace('http://www.w3.org/2002/07/owl#')\n",
    "#XML        = Namespace('http://www.w3.org/XML/1998/namespace')\n",
    "#XSD        = Namespace('http://www.w3.org/2001/XMLSchema#')\n",
    "#DCTERMS    = Namespace('http://purl.org/dc/terms/')\n",
    "\n",
    "#BRICK      = Namespace('https://brickschema.org/schema/1.0.3/Brick#')\n",
    "#BRICKFRAME = Namespace('https://brickschema.org/schema/1.0.3/BrickFrame#')\n",
    "#BRICKTAG   = Namespace('https://brickschema.org/schema/1.0.3/BrickTag#')\n",
    "#BRICKUSE   = Namespace('https://brickschema.org/schema/1.0.3/BrickUse#')\n",
    "\n",
    "#MLS        = Namespace('http://www.w3.org/ns/mls#')\n",
    "\n",
    "#NS         = Namespace('http://inter-space.org/learning_model#')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Ontology:\n",
    "\n",
    "    # prefixes\n",
    "    NS = {\n",
    "        'rdf'        : Namespace('http://www.w3.org/1999/02/22-rdf-syntax-ns#'),\n",
    "        'rdfs'       : Namespace('http://www.w3.org/2000/01/rdf-schema#'),\n",
    "        'owl'        : Namespace('http://www.w3.org/2002/07/owl#'),\n",
    "        'xml'        : Namespace('http://www.w3.org/XML/1998/namespace'),\n",
    "        'xsd'        : Namespace('http://www.w3.org/2001/XMLSchema#'),\n",
    "        'dcterms'    : Namespace('http://purl.org/dc/terms/'),\n",
    "        'brick'      : Namespace('https://brickschema.org/schema/1.0.3/Brick#'),\n",
    "        'brickframe' : Namespace('https://brickschema.org/schema/1.0.3/BrickFrame#'),\n",
    "        'bricktag'   : Namespace('https://brickschema.org/schema/1.0.3/BrickTag#'),\n",
    "        'brickuse'   : Namespace('https://brickschema.org/schema/1.0.3/BrickUse#'),\n",
    "        'mls'        : Namespace('http://www.w3.org/ns/mls#'),\n",
    "        'isl'        : Namespace('http://inter-space.org/learning_model#')\n",
    "    }\n",
    "    \n",
    "    \n",
    "    \n",
    "    metrics = {\n",
    "        'mae': 'MeanAbsoluteError',\n",
    "        'mean_absolute_error': 'MeanAbsoluteError',\n",
    "\n",
    "        'mse': 'MeanSquaredError',\n",
    "        'mean_squared_error': 'MeanSquaredError'\n",
    "    }\n",
    "\n",
    "    \n",
    "    def __init__(self, conjunctive = True):\n",
    "        \n",
    "        if conjunctive: \n",
    "            # create conjunctive graph for NQuads serialization\n",
    "            self.g = ConjunctiveGraph()\n",
    "        else:\n",
    "            self.g = Graph()\n",
    "\n",
    "        for prefix in self.NS:\n",
    "            self.g.bind(prefix, self.NS[prefix])\n",
    "        \n",
    "        # keep track of trained models and their corresponding data splits\n",
    "        self.models = {}\n",
    "            \n",
    "        \n",
    "    def load_ttl(self, ttl_file):\n",
    "        self.g.parse(ttl_file, format=\"turtle\")\n",
    "          \n",
    "            \n",
    "    def get_uuid(self):\n",
    "        return uuid.uuid4().hex\n",
    "    \n",
    "    \n",
    "    def get_building(self):\n",
    "        return next(self.g.subjects(self.NS['rdf'].type, self.NS['brick'].Building))\n",
    "    \n",
    "    \n",
    "    def add_space(self, room_name, statements):\n",
    "        space = self.NS['isl'][room_name]\n",
    "        self.g.add((space, RDF.type, self.NS['brick'].Room))\n",
    "        \n",
    "        for p,o in statements:\n",
    "            self.g.add((space, self.entity(p), self.entity(o)))\n",
    "        \n",
    "        building = self.get_building()\n",
    "        self.g.add((building, self.NS['brickframe'].Contains, space))\n",
    "        return space\n",
    "        \n",
    "        \n",
    "    def add_sensor(self, sensor_name, statements):\n",
    "        sensor = self.NS['isl'][sensor_name]\n",
    "        \n",
    "        for p,o in statements:\n",
    "            self.g.add((sensor, self.entity(p), self.entity(o)))\n",
    "        \n",
    "        \n",
    "        \n",
    "    \n",
    "    \"\"\" \n",
    "    Run is an ML flow that applies a model in a specific dataset to achive the given task.\n",
    "    If it is a trainig run, there wis no input model. Input model is for evaluation and \n",
    "    test purposes. \n",
    "        - test_metrics: metrics of the test run. For adding a trained model should be None\n",
    "        - built_on: in case of transfer learning, adds the original model\n",
    "        \n",
    "    TODO: remove evaluations for training phase\n",
    "        \n",
    "    \"\"\"    \n",
    "    def add_run(self,  model_filename, task_name, split_id, test_metrics = None, built_on = None):\n",
    "        \n",
    "        #train_run = split_id.endswith('/train')\n",
    "        train_run = test_metrics is None\n",
    "        \n",
    "        # tensorflow model\n",
    "        tf_model = tf.keras.models.load_model(model_filename)\n",
    "        \n",
    "        # model\n",
    "        model_id = self.__extract_id(model_filename)\n",
    "        model = self.NS['isl'][model_id]\n",
    "        run_id = 'run_' + self.get_uuid()\n",
    "        ds_id = split_id.split('/')[0]\n",
    "        \n",
    "        model_run = self.NS['isl'][run_id]\n",
    "        self.g.add((model_run, self.NS['rdf'].type, self.NS['mls'].Run))\n",
    "        \n",
    "        \n",
    "        \n",
    "        learning_task = None\n",
    "        \n",
    "        if train_run:\n",
    "            \n",
    "            self.models[model_id] = {'dataset': ds_id, 'built_on': built_on}\n",
    "            \n",
    "            self.g.add((model, self.NS['rdf'].type, self.NS['mls'].Model))\n",
    "            self.g.add((model, self.NS['dcterms'].identifier, Literal(model_filename)))\n",
    "            self.g.add((model_run, self.NS['mls'].hasOutput, model))\n",
    "            \n",
    "            if built_on:\n",
    "                self.g.add((model, self.NS['mls'].builtOn, self.NS['isl']['model_'+built_on]))\n",
    "                learning_task = self.NS['mls'].TransferLearning\n",
    "            else:\n",
    "                learning_task = self.NS['mls'].Training\n",
    "                \n",
    "        else:\n",
    "\n",
    "            self.g.add((model_run, self.NS['mls'].hasInput, model))\n",
    "            test_run = split_id.endswith('/test')\n",
    "\n",
    "            # default learning task\n",
    "            learning_task = self.NS['mls'].ReuseEvaluation\n",
    "            \n",
    "            if model_id in self.models:\n",
    "                \n",
    "                if ds_id == self.models[model_id]['dataset']:\n",
    "                    # evaluate on the same dataset as training \n",
    "                    if self.models[model_id]['built_on']:\n",
    "                        learning_task = self.NS['mls'].TransferEvaluation\n",
    "                    else:\n",
    "                        learning_task = self.NS['mls'].TrainingEvaluation\n",
    "                    \n",
    "            \"\"\"\n",
    "            keep track of learning runs and corresponding datasets\n",
    "\n",
    "            Training evaluation (Training Evaluation)\n",
    "               - trained on the test split of same dataset\n",
    "\n",
    "            Reusing evaluation (Reuse Evaluation)\n",
    "               - different dataset \n",
    "\n",
    "            Fine-tuning evaluation (Transfer Evaluation)\n",
    "            Repurposing evaluation (Transfer Evaluation)\n",
    "               - trained on finetune split of same dataset\n",
    "            \"\"\"\n",
    "\n",
    "            \n",
    "            # model evaluations\n",
    "            eval_id = run_id + '/eval'\n",
    "            model_evaluation = self.NS['isl'][eval_id]\n",
    "            self.g.add((model_evaluation, self.NS['rdf'].type, self.NS['mls'].ModelEvaluation))\n",
    "\n",
    "\n",
    "            # metrics will be added for test runs\n",
    "            for m in tf_model.metrics:\n",
    "                if m.name in self.metrics:\n",
    "                    metric_name = self.metrics[m.name]\n",
    "                    metric_class = self.NS['mls'][metric_name]\n",
    "\n",
    "                    metric = self.NS['isl'][eval_id + '/' + m.name]\n",
    "                    self.g.add((metric, self.NS['rdf'].type, metric_class))\n",
    "                    self.g.add((metric, self.NS['mls'].hasValue, Literal(test_metrics[m.name], datatype=self.NS['xsd'].float) ))\n",
    "                    self.g.add((model_evaluation, self.NS['mls'].specifiedBy, metric))\n",
    "                    \n",
    "            self.g.add((model_run, self.NS['mls'].hasOutput, model_evaluation))\n",
    "\n",
    "        \n",
    "        # task\n",
    "        task = self.NS['isl'][task_name]\n",
    "        self.g.add((model_run, self.NS['mls'].achieves, task))\n",
    "        \n",
    "        self.g.add((model_run, self.NS['mls'].hasInput, self.NS['isl'][split_id]))\n",
    "        self.g.add((model_run, self.NS['mls'].executes, self.NS['isl']['tensorflow']))\n",
    "        \n",
    "        backend = type(tf_model).__module__+'.'+type(tf_model).__name__\n",
    "        self.g.add((model_run, self.NS['mls'].implements, Literal(backend)))\n",
    "        \n",
    "        self.g.add((model_run, self.NS['mls'].realizes, learning_task ))\n",
    "        \n",
    "        \n",
    "        return model_run\n",
    "        \n",
    "    \n",
    "    \"\"\"\n",
    "    Receive the dataset as dataframe and add it to ontology. \n",
    "    Also the references to BRICK entities (e.g. datatype and sensors) can be added.\n",
    "    \"\"\"\n",
    "    def add_dataset(self, ds_id, brick_refs=None):\n",
    "        \n",
    "        stats_dic = {\n",
    "            'min': {'class': self.NS['mls'].MinValue, 'datatype': self.NS['xsd'].float },\n",
    "            'max': {'class': self.NS['mls'].MaxValue, 'datatype': self.NS['xsd'].float },\n",
    "            'mean': {'class': self.NS['mls'].MeanValue, 'datatype': self.NS['xsd'].float },\n",
    "            'std': {'class': self.NS['mls'].StdValue, 'datatype': self.NS['xsd'].float }\n",
    "        }\n",
    "        \n",
    "        split_dict = {\n",
    "            'test' : 'TestDataset',\n",
    "            'train': 'TrainDataset',\n",
    "            'finetune': 'FineTuningDataset',\n",
    "            'validation': 'ValidationDataset'\n",
    "        }\n",
    "\n",
    "        dataset_filename = 'dataset/ds_{}.h5'.format(ds_id)\n",
    "        \n",
    "        ds = pd.HDFStore(dataset_filename, 'r')\n",
    "        split_keys = [k[1:] for k in ds.keys()]\n",
    "        \n",
    "        # get structure from the first data split\n",
    "        key0 = split_keys[0]\n",
    "\n",
    "        dataset_uri = self.NS['isl'][ds_id]\n",
    "        self.g.add((dataset_uri, self.NS['rdf'].type, self.NS['mls'].Dataset))\n",
    "        \n",
    "        \n",
    "        for col in ds[key0].columns:\n",
    "            \n",
    "            feature = self.NS['isl'][ds_id+'/'+col]\n",
    "            self.g.add((feature, self.NS['rdf'].type, self.NS['mls'].Feature))\n",
    "            self.g.add((dataset_uri, self.NS['mls'].hasPart, feature))\n",
    "\n",
    "            \n",
    "            # add brick references if any\n",
    "            if brick_refs and col in brick_refs:\n",
    "                statements = brick_refs[col]\n",
    "                for p,o in statements:\n",
    "                    self.g.add((feature, self.entity(p), self.entity(o)))\n",
    "                \n",
    "\n",
    "        \n",
    "        # add dataset splits    \n",
    "        for key in split_keys:\n",
    "            # add split definition\n",
    "            split_id = ds_id + '/' + key\n",
    "            split = self.NS['isl'][split_id]\n",
    "            split_class = split_dict[key]\n",
    "            self.g.add((split, self.NS['rdf'].type, self.NS['mls'][split_class]))\n",
    "            self.g.add((dataset_uri, self.NS['mls'].hasSplit, split))\n",
    "            \n",
    "           \n",
    "            # add number of instances to dataset split\n",
    "            rows, _ = ds[key].shape\n",
    "            val = self.NS['isl'][split_id+'/'+'count']\n",
    "            self.g.add((val, self.NS['rdf'].type, self.NS['mls'].DatasetCharacteristic))\n",
    "            self.g.add((val, self.NS['mls'].hasValue, Literal(rows, datatype=self.NS['xsd'].long) ))\n",
    "            self.g.add((split, self.NS['mls'].hasQuality, val))\n",
    "\n",
    "        \n",
    "            stats = ds[key].describe()\n",
    "            \n",
    "            for col in ds[key].columns:\n",
    "                column = self.NS['isl'][split_id+'/'+col]\n",
    "                feature = self.NS['isl'][ds_id+'/'+col]\n",
    "                \n",
    "                self.g.add((column, self.NS['rdf'].type, self.NS['mls'].Column))\n",
    "                self.g.add((column, self.NS['mls'].specifiedBy, feature))\n",
    "\n",
    "\n",
    "                for ind in stats_dic: # stat indicators\n",
    "\n",
    "                    # quality indicators\n",
    "                    val = self.NS['isl'][split_id+'/'+col+'/'+ind]\n",
    "                    self.g.add((val, self.NS['rdf'].type, stats_dic[ind]['class']))\n",
    "                    self.g.add((val, self.NS['mls'].hasValue, Literal(stats[col][ind], datatype=stats_dic[ind]['datatype']) ))\n",
    "\n",
    "                    self.g.add((column, self.NS['mls'].hasQuality, val))\n",
    "                    \n",
    "        ds.close()\n",
    "        return\n",
    "\n",
    "\n",
    "    \"\"\" extract entity id from filename\n",
    "    \"\"\"\n",
    "    def __extract_id(self, filename):\n",
    "        start = filename.rfind('/') + 1\n",
    "        end = filename.rfind('.')\n",
    "        return filename[start:end]\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "    \"\"\" Geneartes a qualified entity for the given name with prefix\n",
    "    \"\"\"\n",
    "    def entity(self, name):\n",
    "        if type(name) in [Literal, URIRef]:\n",
    "            return name\n",
    "        elif ':' in name:\n",
    "            prefix, label = name.split(':')\n",
    "            return self.NS[prefix][label]\n",
    "        else:\n",
    "            return Literal(name, datatype=self.NS['xsd'].string)\n",
    "        \n",
    "        \n",
    "    def serialize(self):\n",
    "        print(self.g.serialize(format='n3').decode(\"utf-8\"))\n",
    "        \n",
    "    def save(self, file_name):\n",
    "        formats = {\n",
    "            'ttl': 'turtle',\n",
    "            'nq':'nquads'\n",
    "        }\n",
    "        _, ext = file_name.split('.')\n",
    "        \n",
    "        if ext == 'ttl':\n",
    "            self.g.serialize(destination='ontology/'+file_name, format=formats[ext])\n",
    "            \n",
    "        elif ext =='nq':\n",
    "            # shorten namespaces and remove graph part\n",
    "            st = self.g.serialize(format=formats[ext]).decode(\"utf-8\")\n",
    "\n",
    "            for title in self.NS:\n",
    "                st = st.replace(str(self.NS[title]), title+':')\n",
    "\n",
    "            lines = st.split('\\n')\n",
    "\n",
    "            with open('ontology/'+file_name, 'w') as f:\n",
    "                for line in lines:\n",
    "                    elms = line.split(' ')\n",
    "                    cnt = len(elms)\n",
    "                    if cnt  > 2:\n",
    "                        f.write(' '.join(elms[:cnt-2]) + ' .\\n')        \n",
    "                        \n",
    "        else:\n",
    "            print('Unknown file extension!')\n",
    "            \n",
    "\n",
    "            \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 471,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont = Ontology()\n",
    "ont.load_ttl('ontology/base_graph.ttl')\n",
    "#ont.add_dataset(train_dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 472,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 167.9990 - mean_absolute_error: 9.7587 - mean_squared_error: 167.9990\n",
      "4/4 - 0s - loss: 63.6485 - mean_absolute_error: 5.4978 - mean_squared_error: 63.6485\n",
      "4/4 - 0s - loss: 129.4320 - mean_absolute_error: 9.6490 - mean_squared_error: 129.4320\n",
      "4/4 - 0s - loss: 327.7086 - mean_absolute_error: 13.4043 - mean_squared_error: 327.7086\n",
      "4/4 - 0s - loss: 21.8347 - mean_absolute_error: 3.7022 - mean_squared_error: 21.8347\n",
      "4/4 - 0s - loss: 65.6372 - mean_absolute_error: 6.6769 - mean_squared_error: 65.6372\n",
      "4/4 - 0s - loss: 328.2416 - mean_absolute_error: 13.3242 - mean_squared_error: 328.2416\n",
      "4/4 - 0s - loss: 34.1794 - mean_absolute_error: 4.6656 - mean_squared_error: 34.1794\n",
      "4/4 - 0s - loss: 40.5407 - mean_absolute_error: 5.0176 - mean_squared_error: 40.5407\n"
     ]
    }
   ],
   "source": [
    "for room_id, room in rooms.items():\n",
    "    \n",
    "    #dataset = room_dataset[room_id].copy()\n",
    "    #train_dataset = dataset.sample(frac=0.8,random_state=0)\n",
    "    #test_dataset = dataset.drop(train_dataset.index)\n",
    "    \n",
    "    #ds_uuid = uuid.uuid4().hex\n",
    "    #filename = 'dataset/ds_{}.h5'.format(ds_uuid)\n",
    "    #store = pd.HDFStore(filename)\n",
    "    #store['train'] = train_dataset\n",
    "    #store['test'] = test_dataset\n",
    "    #store.close()\n",
    "    \n",
    "    \n",
    "    brick_ref = {\n",
    "        'co2': [\n",
    "            ('rdf:type', 'brick:CO2_Level'), \n",
    "            ('rdfs:label', 'Room CO2 Level'),\n",
    "            ('brickframe:isMeasuredBy', 'isl:/co2_sensors/'+str(room_id))\n",
    "        ],\n",
    "        'occupancy': [\n",
    "            ('rdf:type', 'brick:Occupancy'),\n",
    "            ('rdfs:label', 'Room Occupancy'),\n",
    "            ('brickframe:isMeasuredBy', 'isl:/occ_sensors/'+str(room_id))\n",
    "        ] \n",
    "    }    \n",
    "    \n",
    "\n",
    "    \n",
    "    ont.add_dataset(room['dataset_uuid'], brick_ref)\n",
    "    \n",
    "    #model_uuid = uuid.uuid4().hex\n",
    "    #model_filename = 'model/model_{}.h5'.format(model_uuid)\n",
    "    #rooms[room_id]['model'].save(model_filename)\n",
    "    \n",
    "    # add train run\n",
    "    train_split_id = room['dataset_uuid'] + '/train'\n",
    "    ont.add_run(room['model_file'], 'occupancy_task', train_split_id)\n",
    "\n",
    "\n",
    "# add test runs (source room model + target room data)\n",
    "for source_room in rooms:\n",
    "    \n",
    "    source_model = rooms[source_room]['model']\n",
    "\n",
    "    for target_room in rooms:\n",
    "        \n",
    "        metric_values = source_model.evaluate(rooms[target_room]['test_data'], rooms[target_room]['test_labels'], verbose=2)\n",
    "        metric_names = [m.name for m in source_model.metrics]\n",
    "        metrics = dict(zip(metric_names, metric_values))\n",
    "\n",
    "        test_split_id = rooms[target_room]['dataset_uuid'] + '/test'\n",
    "        ont.add_run(rooms[source_room]['model_file'], 'occupancy_task', test_split_id, metrics)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 308,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Adding Amin's Room\n",
    "space_info = [ \n",
    "    ('isl:hasArea', Literal('12', datatype=XSD.float)), \n",
    "    ('isl:hasUsage', 'Office')]\n",
    "room = ont.add_space('/rooms/amin', space_info)\n",
    "\n",
    "\n",
    "sensor_info = [\n",
    "    ('brickframe:isPointOf', room), \n",
    "    ('rdf:type', 'brick:Luminance_Sensor'), \n",
    "    ('brickframe:hasMeasurement', Literal('ws://192.168.0.115:9001/light', datatype=XSD.string))] \n",
    "ont.add_sensor('/luminance_sensors/cpx_l', sensor_info)\n",
    "\n",
    "     \n",
    "sensor_info = [\n",
    "    ('brickframe:isPointOf', room), \n",
    "    ('rdf:type', 'brick:Temperature_Sensor'), \n",
    "    ('brickframe:hasMeasurement', Literal('ws://192.168.0.115:9001/temperature', datatype=XSD.string))] \n",
    "ont.add_sensor('/temperature_sensors/cpx_t', sensor_info)\n",
    "     \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.serialize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 407,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.nq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 478,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.ttl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ds_id = '0d9a0f922a814f3298ee73b8b9c130e6'\n",
    "room3_ds_id = 'ce357a6ce1ee45d1bc58596222f828a3'\n",
    "dataset_filename = f'dataset/ds_{room3_ds_id}.h5'\n",
    "room3_ds = pd.HDFStore(dataset_filename, 'r')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [],
   "source": [
    "room3_model_id = 'a52818ed338744f2a6d9026f821f324e'\n",
    "model_filename = f'model/model_{room3_model_id}.h5'\n",
    "room3_model = tf.keras.models.load_model(model_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_32\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "normalization_29 (Normalizat (None, 1)                 3         \n",
      "_________________________________________________________________\n",
      "dense_96 (Dense)             (None, 64)                128       \n",
      "_________________________________________________________________\n",
      "dense_97 (Dense)             (None, 64)                4160      \n",
      "_________________________________________________________________\n",
      "dense_98 (Dense)             (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 4,356\n",
      "Trainable params: 4,353\n",
      "Non-trainable params: 3\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "tf_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_predictions = room3_model.predict(rooms[room_id]['test_data']).flatten()\n",
    "\n",
    "a = plt.axes(aspect='equal')\n",
    "plt.scatter(test_labels, test_predictions)\n",
    "plt.xlabel('True Values [Occupants]')\n",
    "plt.ylabel('Predictions [Occupants]')\n",
    "lims = [0, 80]\n",
    "plt.xlim(lims)\n",
    "plt.ylim(lims)\n",
    "_ = plt.plot(lims, lims)\n",
    "\n",
    "error = test_predictions - rooms[room_id]['test_labels']\n",
    "plt.hist(error, bins = 25)\n",
    "plt.xlabel(\"Prediction Error [Occupants]\")\n",
    "_ = plt.ylabel(\"Count\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3/3 - 0s - loss: 44.2590 - mean_absolute_error: 5.3428 - mean_squared_error: 44.2590\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 44.25895309448242,\n",
       " 'mean_absolute_error': 5.342784404754639,\n",
       " 'mean_squared_error': 44.25895309448242}"
      ]
     },
     "execution_count": 145,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metric_values = room3_model.evaluate(test_dataset['co2'], test_dataset['occupancy'], verbose=2)\n",
    "metric_names = [m.name for m in room3_model.metrics]\n",
    "metrics = dict(zip(metric_names, metric_values))\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:5 out of the last 22 calls to <function Model.make_predict_function.<locals>.predict_function at 0x7fb2fc0e88c0> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/tutorials/customization/performance#python_or_tensor_args and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_predictions = room3_model.predict(test_dataset['co2']).flatten()\n",
    "error = test_predictions - test_dataset['occupancy']\n",
    "plt.hist(error, bins = 25)\n",
    "plt.xlabel(\"Prediction Error [Occupants]\")\n",
    "_ = plt.ylabel(\"Count\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_predictions = new_model.predict(test_dataset['co2']).flatten()\n",
    "error = test_predictions - test_dataset['occupancy']\n",
    "plt.hist(error, bins = 25)\n",
    "plt.xlabel(\"Prediction Error [Occupants]\")\n",
    "_ = plt.ylabel(\"Count\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Retraing dataset added:  dataset/ds_aadd267319fe464286c4e70b68191924.h5\n"
     ]
    }
   ],
   "source": [
    "# TO BE DELETED\n",
    "# retrain on room 2 data\n",
    "dataset = room_dataset[2].copy()\n",
    "\n",
    "#retrain_dataset = dataset.sample(frac=0.1,random_state=0)\n",
    "retrain_dataset = dataset.sample(n=100,random_state=0)\n",
    "retrain_labels = retrain_dataset.pop('occupancy')\n",
    "\n",
    "rest_dataset = dataset.drop(retrain_dataset.index)\n",
    "\n",
    "#test_dataset = rest_dataset.sample(frac=0.1,random_state=0)\n",
    "test_dataset = rest_dataset.sample(n=100,random_state=0)\n",
    "\n",
    "\n",
    "ds_uuid = uuid.uuid4().hex\n",
    "ds_file = f'dataset/ds_{ds_uuid}.h5'\n",
    "store = pd.HDFStore(ds_file)\n",
    "store['train'] = retrain_dataset\n",
    "store['test'] = test_dataset\n",
    "store.close()\n",
    "print('Retraing dataset added: ', ds_file)\n",
    "\n",
    "ont.add_dataset(ds_uuid, brick_ref)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transfer Learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 474,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Take the model from room 2\n",
    "model_room_id = 2\n",
    "\n",
    "orig_model = rooms[model_room_id]['model']\n",
    "\n",
    "# load again to repurpose\n",
    "orig_model_id = rooms[model_room_id]['model_uuid']\n",
    "orig_model_filename = f'model/model_{orig_model_id}.h5'\n",
    "\n",
    "new_model = tf.keras.models.load_model(orig_model_filename)\n",
    "\n",
    "new_model.compile(\n",
    "    loss='mse',\n",
    "    optimizer=tf.keras.optimizers.Adam(0.00001),\n",
    "    #optimizer=keras.optimizers.Adam(1e-5),  # Low learning rate\n",
    "    # use complete metric names\n",
    "    metrics=['mean_absolute_error', 'mean_squared_error'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 475,
   "metadata": {},
   "outputs": [],
   "source": [
    "# repurpose for room 3\n",
    "ds_room_id = 3\n",
    "ds_id = rooms[ds_room_id]['dataset_uuid']\n",
    "dataset_filename = f'dataset/ds_{ds_id}.h5'\n",
    "room_ds = pd.HDFStore(dataset_filename, 'r')\n",
    "\n",
    "finetune_dataset = room_ds['finetune'] \n",
    "finetune_labels = finetune_dataset.pop('occupancy')\n",
    "\n",
    "test_dataset = room_ds['test'] \n",
    "room_ds.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 476,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Epoch: 0, loss:69.2057,  mean_absolute_error:7.2603,  mean_squared_error:69.2057,  val_loss:44.3335,  val_mean_absolute_error:5.5014,  val_mean_squared_error:44.3335,  \n",
      "....................................................................................................\n",
      "Epoch: 100, loss:63.7513,  mean_absolute_error:6.9454,  mean_squared_error:63.7513,  val_loss:41.2549,  val_mean_absolute_error:5.2262,  val_mean_squared_error:41.2549,  \n",
      "....................................................................................................\n",
      "Epoch: 200, loss:59.9275,  mean_absolute_error:6.6961,  mean_squared_error:59.9275,  val_loss:39.1099,  val_mean_absolute_error:5.0424,  val_mean_squared_error:39.1099,  \n",
      "....................................................................................................\n",
      "Epoch: 300, loss:57.2641,  mean_absolute_error:6.4981,  mean_squared_error:57.2641,  val_loss:37.6210,  val_mean_absolute_error:4.9014,  val_mean_squared_error:37.6210,  \n",
      "....................................................................................................\n",
      "Epoch: 400, loss:55.4976,  mean_absolute_error:6.3349,  mean_squared_error:55.4976,  val_loss:36.6248,  val_mean_absolute_error:4.8150,  val_mean_squared_error:36.6248,  \n",
      "....................................................................................................\n",
      "Epoch: 500, loss:54.2678,  mean_absolute_error:6.2005,  mean_squared_error:54.2678,  val_loss:35.9127,  val_mean_absolute_error:4.7804,  val_mean_squared_error:35.9127,  \n",
      "....................................................................................................\n",
      "Epoch: 600, loss:53.3908,  mean_absolute_error:6.1021,  mean_squared_error:53.3908,  val_loss:35.3935,  val_mean_absolute_error:4.7565,  val_mean_squared_error:35.3935,  \n",
      "....................................................................................................\n",
      "Epoch: 700, loss:52.7617,  mean_absolute_error:6.0204,  mean_squared_error:52.7617,  val_loss:35.0007,  val_mean_absolute_error:4.7395,  val_mean_squared_error:35.0007,  \n",
      "....................................................................................................\n",
      "Epoch: 800, loss:52.2838,  mean_absolute_error:5.9535,  mean_squared_error:52.2838,  val_loss:34.6844,  val_mean_absolute_error:4.7237,  val_mean_squared_error:34.6844,  \n",
      "....................................................................................................\n",
      "Epoch: 900, loss:51.9192,  mean_absolute_error:5.9083,  mean_squared_error:51.9192,  val_loss:34.4348,  val_mean_absolute_error:4.7105,  val_mean_squared_error:34.4348,  \n",
      "...................................................................................................."
     ]
    }
   ],
   "source": [
    "EPOCHS = 1000\n",
    "\n",
    "# The patience parameter is the amount of epochs to check for improvement\n",
    "early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)\n",
    "\n",
    "history = new_model.fit(finetune_dataset, finetune_labels, \n",
    "                    epochs=EPOCHS, validation_split = 0.2, verbose=0, \n",
    "                    callbacks=[early_stop, tfdocs.modeling.EpochDots()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 477,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "new model_id: b4895793546a4048b0656859e1430349\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "rdflib.term.URIRef('http://inter-space.org/learning_model#run_33c8c6b8ddfe42c6b5bb9a76259a2d96')"
      ]
     },
     "execution_count": 477,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# store repurposed model\n",
    "new_model_uuid = uuid.uuid4().hex\n",
    "print('new model_id:', new_model_uuid)\n",
    "\n",
    "new_model_file = f'model/model_{new_model_uuid}.h5'\n",
    "new_model.save(new_model_file)\n",
    "\n",
    "train_split_id = ds_id + '/finetune'\n",
    "\n",
    "# add training run\n",
    "ont.add_run(new_model_file, 'occupancy_task', train_split_id, built_on = orig_model_id)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 483,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 52.9462 - mean_absolute_error: 5.6570 - mean_squared_error: 52.9462\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "rdflib.term.URIRef('http://inter-space.org/learning_model#run_81a2068ffb1f4379bf45fd7d2212cc93')"
      ]
     },
     "execution_count": 483,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# add repurposing run to ontology\n",
    "metric_values = new_model.evaluate(test_dataset['co2'], test_dataset['occupancy'], verbose=2)\n",
    "metric_names = [m.name for m in new_model.metrics]\n",
    "metrics = dict(zip(metric_names, metric_values))\n",
    "metrics\n",
    "\n",
    "test_split_id = ds_uuid + '/test'\n",
    "ont.add_run(new_model_file, 'occupancy_task', test_split_id, metrics)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 484,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.ttl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 398,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.nq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 52.9176 - mean_absolute_error: 5.6542 - mean_squared_error: 52.9176\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 52.917598724365234,\n",
       " 'mean_absolute_error': 5.654246807098389,\n",
       " 'mean_squared_error': 52.917598724365234}"
      ]
     },
     "execution_count": 316,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metric_values = new_model.evaluate(test_dataset['co2'], test_dataset['occupancy'], verbose=2)\n",
    "metric_names = [m.name for m in new_model.metrics]\n",
    "metrics = dict(zip(metric_names, metric_values))\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 304,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 65.6372 - mean_absolute_error: 6.6769 - mean_squared_error: 65.6372\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 65.63717651367188,\n",
       " 'mean_absolute_error': 6.6768574714660645,\n",
       " 'mean_squared_error': 65.63717651367188}"
      ]
     },
     "execution_count": 304,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metric_values = orig_model.evaluate(test_dataset['co2'], test_dataset['occupancy'], verbose=2)\n",
    "metric_names = [m.name for m in room3_model.metrics]\n",
    "metrics = dict(zip(metric_names, metric_values))\n",
    "metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "29.372221\n",
      "3.9798098\n",
      "29.372221\n"
     ]
    }
   ],
   "source": [
    "for m in new_model.metrics:\n",
    "    print(m.result().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.0\n",
      "0.0\n",
      "0.0\n"
     ]
    }
   ],
   "source": [
    "for m in room3_model.metrics:\n",
    "    print(m.result().numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_model = tf.keras.models.clone_model(room3_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 22.1342 - mean_absolute_error: 3.7362 - mean_squared_error: 22.1342\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss, mae, mse = rooms[2]['model'].evaluate(rooms[2]['test_data'], rooms[2]['test_labels'], verbose=2)\n",
    "test_predictions = rooms[2]['model'].predict(rooms[2]['test_data']).flatten()\n",
    "error = test_predictions - rooms[2]['test_labels'] \n",
    "data = plt.hist(error, bins = 25)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-13.45270538, -12.50981117, -11.56691696, -10.62402275,\n",
       "        -9.68112854,  -8.73823433,  -7.79534012,  -6.85244591,\n",
       "        -5.9095517 ,  -4.96665749,  -4.02376328,  -3.08086906,\n",
       "        -2.13797485,  -1.19508064,  -0.25218643,   0.69070778,\n",
       "         1.63360199,   2.5764962 ,   3.51939041,   4.46228462,\n",
       "         5.40517883,   6.34807304,   7.29096725,   8.23386147,\n",
       "         9.17675568,  10.11964989])"
      ]
     },
     "execution_count": 272,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 329,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'http://www.w3.org/1999/02/22-rdf-syntax-ns#'"
      ]
     },
     "execution_count": 329,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "str(ont.NS['rdf'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 378,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'<isl:/damper_sensors/2> <rdf:type> \"test file\" <file:///> '"
      ]
     },
     "execution_count": 378,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "st = '<isl:/damper_sensors/2> <rdf:type> \"test file\" <file:///> .'\n",
    "post = st.rfind(' ', 2)\n",
    "st[:pos]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 379,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['<isl:/damper_sensors/2>', '<rdf:type>', '\"test', 'file\"', '<file:///>', '.']"
      ]
     },
     "execution_count": 379,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "st.split(' ')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 385,
   "metadata": {},
   "outputs": [],
   "source": [
    "st = ont.g.serialize(format='nquads').decode(\"utf-8\")\n",
    "lines = st.split('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 389,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<http://inter-space.org/learning_model#/damper_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#hasCapacity> <http://www.w3.org/2000/01/rdf-schema#label> \"Seating capacity of room\"\n",
      "<http://inter-space.org/learning_model#/damper_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV_Damper_Position_Sensor>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#temperature_room_2.csv>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#illuminance_room_1.csv>\n",
      "<http://inter-space.org/learning_model#/rooms/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#/rooms/3> <http://inter-space.org/learning_model#hasVolume> \"418.75\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://www.w3.org/ns/mls#MeanValue> <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.w3.org/ns/mls#FeatureCharacteristic>\n",
      "<http://inter-space.org/learning_model#/rooms/3> <http://inter-space.org/learning_model#hasUsage> \"study\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#/rooms/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#temperature_room_1.csv>\n",
      "<http://inter-space.org/learning_model#/rooms/2> <http://inter-space.org/learning_model#hasCapacity> \"32\"^^<http://www.w3.org/2001/XMLSchema#int>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Temperature_Sensor>\n",
      "<http://inter-space.org/learning_model#hasUsage> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#DatatypeProperty>\n",
      "<http://inter-space.org/learning_model#hasVolume> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#DatatypeProperty>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#humidity_room_1.csv>\n",
      "<http://www.w3.org/ns/mls#MinValue> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#Class>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#illuminance_room_3.csv>\n",
      "<http://www.w3.org/ns/mls#StdValue> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#Class>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#vav_room_1.csv>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#temperature_room_3.csv>\n",
      "<http://inter-space.org/learning_model#hasCapacity> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#DatatypeProperty>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Humidity_Sensor>\n",
      "<http://inter-space.org/learning_model#/ahus/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#AHU>\n",
      "<http://inter-space.org/learning_model#/rooms/1> <http://inter-space.org/learning_model#hasArea> \"139.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy_Sensor>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Illuminance_Sensor>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#hasUsage> <http://www.w3.org/2000/01/rdf-schema#domain> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#hasArea> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#DatatypeProperty>\n",
      "<http://inter-space.org/learning_model#/vavs/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV>\n",
      "<http://inter-space.org/learning_model#/rooms/1> <http://inter-space.org/learning_model#hasCapacity> \"84\"^^<http://www.w3.org/2001/XMLSchema#int>\n",
      "<http://inter-space.org/learning_model#hasUsage> <http://www.w3.org/2000/01/rdf-schema#label> \"Space usage type\"\n",
      "<http://inter-space.org/learning_model#/co2_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#co2_room_1.csv>\n",
      "<http://inter-space.org/learning_model#/rooms/2> <http://inter-space.org/learning_model#hasArea> \"125.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#co2_room_2.csv>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Illuminance_Sensor>\n",
      "<http://www.w3.org/ns/mls#MeanValue> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#Class>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Sensor>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPartOf> <http://inter-space.org/learning_model#/vavs/2>\n",
      "<http://inter-space.org/learning_model#/ahus/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/vavs/2>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#humidity_room_2.csv>\n",
      "<http://inter-space.org/learning_model#hasArea> <http://www.w3.org/2000/01/rdf-schema#label> \"Area size in square meters\"\n",
      "<http://inter-space.org/learning_model#/co2_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Sensor>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Temperature_Sensor>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#hasVolume> <http://www.w3.org/2000/01/rdf-schema#label> \"Volume in cubic meters\"\n",
      "<http://inter-space.org/learning_model#/vavs/3> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV_Damper_Position_Sensor>\n",
      "<http://inter-space.org/learning_model#/rooms/1> <http://inter-space.org/learning_model#hasVolume> \"461.48\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Humidity_Sensor>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#occupant_count_room_1.csv>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Humidity_Sensor>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#vav_room_3.csv>\n",
      "<http://inter-space.org/learning_model#/vavs/2> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#co2_room_3.csv>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#hasVolume> <http://www.w3.org/2000/01/rdf-schema#domain> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#/buildings/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Contains> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/rooms/3> <http://inter-space.org/learning_model#hasCapacity> \"32\"^^<http://www.w3.org/2001/XMLSchema#int>\n",
      "<http://inter-space.org/learning_model#/vavs/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPartOf> <http://inter-space.org/learning_model#/vavs/1>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#vav_room_2.csv>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#humidity_room_3.csv>\n",
      "<http://inter-space.org/learning_model#/vavs/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Illuminance_Sensor>\n",
      "<http://inter-space.org/learning_model#/rooms/1> <http://inter-space.org/learning_model#hasUsage> \"lecture\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://www.w3.org/ns/mls#MaxValue> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/2002/07/owl#Class>\n",
      "<http://inter-space.org/learning_model#/rooms/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#/temperature_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Temperature_Sensor>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV_Damper_Position_Sensor>\n",
      "<http://inter-space.org/learning_model#hasArea> <http://www.w3.org/2000/01/rdf-schema#domain> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPartOf> <http://inter-space.org/learning_model#/vavs/3>\n",
      "<http://inter-space.org/learning_model#/rooms/2> <http://inter-space.org/learning_model#hasVolume> \"418.75\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy_Sensor>\n",
      "<http://www.w3.org/ns/mls#MinValue> <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.w3.org/ns/mls#FeatureCharacteristic>\n",
      "<http://www.w3.org/ns/mls#StdValue> <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.w3.org/ns/mls#FeatureCharacteristic>\n",
      "<http://inter-space.org/learning_model#/ahus/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/vavs/1>\n",
      "<http://inter-space.org/learning_model#/ahus/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Feeds> <http://inter-space.org/learning_model#/vavs/3>\n",
      "<http://inter-space.org/learning_model#/buildings/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Contains> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/rooms/3> <http://inter-space.org/learning_model#hasArea> \"125.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#/vavs/2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#VAV>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#occupant_count_room_2.csv>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/buildings/1> <https://brickschema.org/schema/1.0.3/BrickFrame#Contains> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#occupant_count_room_3.csv>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/2>\n",
      "<http://inter-space.org/learning_model#/co2_sensors/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Sensor>\n",
      "<http://inter-space.org/learning_model#/occ_sensors/3> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy_Sensor>\n",
      "<http://inter-space.org/learning_model#/illuminate_sensors/2> <https://brickschema.org/schema/1.0.3/BrickFrame#hasMeasurement> <http://inter-space.org/learning_model#illuminance_room_2.csv>\n",
      "<http://inter-space.org/learning_model#/humidity_sensors/1> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/1>\n",
      "<http://inter-space.org/learning_model#/buildings/1> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Building>\n",
      "<http://inter-space.org/learning_model#/damper_sensors/3> <https://brickschema.org/schema/1.0.3/BrickFrame#isPointOf> <http://inter-space.org/learning_model#/rooms/3>\n",
      "<http://www.w3.org/ns/mls#MaxValue> <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.w3.org/ns/mls#FeatureCharacteristic>\n",
      "<http://inter-space.org/learning_model#/rooms/2> <http://inter-space.org/learning_model#hasUsage> \"study\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#hasCapacity> <http://www.w3.org/2000/01/rdf-schema#domain> <https://brickschema.org/schema/1.0.3/Brick#Room>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/min> <http://www.w3.org/ns/mls#hasValue> \"450.61333719889325\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/max>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"5.017570495605469\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"9.398843930635838\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/min> <http://www.w3.org/ns/mls#hasValue> \"417.67466837565104\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"34.17942810058594\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/max> <http://www.w3.org/ns/mls#hasValue> \"925.3973225911458\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/min> <http://www.w3.org/ns/mls#hasValue> \"451.5733337402344\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy> <http://www.w3.org/2000/01/rdf-schema#label> \"Room Occupancy\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/std>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"328.2415771484375\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"167.99900817871094\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/max>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/count>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TrainDataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"9.648994445800781\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"327.7085876464844\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/min>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/count>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#FineTuningDataset>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"78.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"8.81\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/std>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/max> <http://www.w3.org/ns/mls#hasValue> \"1015.7653442382813\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2> <http://www.w3.org/2000/01/rdf-schema#label> \"Room CO2 Level\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"75.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Dataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/std>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/max>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/std> <http://www.w3.org/ns/mls#hasValue> \"103.5906158965429\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/min>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"9.05\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Dataset>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"64.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TrainDataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2> <http://www.w3.org/2000/01/rdf-schema#label> \"Room CO2 Level\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/count>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/std> <http://www.w3.org/ns/mls#hasValue> \"131.3770682091649\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"13.324183464050293\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"8.03197524041417\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"9.237493526384311\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/mean>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#FineTuningDataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/min> <http://www.w3.org/ns/mls#hasValue> \"425.94133809407555\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"4.665645599365234\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"65.63717651367188\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#FineTuningDataset>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/max> <http://www.w3.org/ns/mls#hasValue> \"1133.9733357747396\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/std>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/mean>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/std>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/occ_sensors/3>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Model>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/max>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"20.32\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"604.6328550923665\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/min> <http://www.w3.org/ns/mls#hasValue> \"453.717328898112\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee> <http://purl.org/dc/terms/identifier> \"model/model_0754017c705548d490065f71ac4364ee.h5\"\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/min> <http://www.w3.org/ns/mls#hasValue> \"425.642670694987\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"3.7022125720977783\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/min>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Level>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/max> <http://www.w3.org/ns/mls#hasValue> \"843.1360087076823\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/min>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/min>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"21.834674835205078\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/min>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"19.985032277970312\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/count> <http://www.w3.org/ns/mls#hasValue> \"692\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Level>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/min>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/count> <http://www.w3.org/ns/mls#hasValue> \"692\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"9.382947976878613\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/max>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/min> <http://www.w3.org/ns/mls#hasValue> \"425.52533671061195\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/mean>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Dataset>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/count>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"560.8944005126954\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"6.796753532816869\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"607.3300352960193\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/std>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7> <http://purl.org/dc/terms/identifier> \"model/model_cac90a7b1bf147bda94625edc5b339f7.h5\"\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"20.721764286650764\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/min>\n",
      "<http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Model>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"43.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"20.65\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/std> <http://www.w3.org/ns/mls#hasValue> \"121.36194756571223\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/max>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/min> <http://www.w3.org/ns/mls#hasValue> \"423.4879994710287\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/max>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/std> <http://www.w3.org/ns/mls#hasValue> \"119.00257089695087\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/min>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/co2_sensors/3>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/std> <http://www.w3.org/ns/mls#hasValue> \"98.23144629765237\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"621.5223477681478\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"8.36\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"55.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/std>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/max>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/max> <http://www.w3.org/ns/mls#hasValue> \"1133.9733357747396\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/mean>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"63.64848709106445\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/std> <http://www.w3.org/ns/mls#hasValue> \"127.91629371799146\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"7.977130733612129\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/max>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"20.43208092485549\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/std>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/max> <http://www.w3.org/ns/mls#hasValue> \"951.8293497721354\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TestDataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/mean>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/std> <http://www.w3.org/ns/mls#hasValue> \"101.45264947004893\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/min>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/min>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"40.540653228759766\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8> <http://www.w3.org/ns/mls#hasPart> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/co2_sensors/3>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/mean>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Model>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/min>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/occ_sensors/3>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"7.035264706814485\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/std> <http://www.w3.org/ns/mls#hasValue> \"128.6885576296004\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Run>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/min>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy> <http://www.w3.org/2000/01/rdf-schema#label> \"Room Occupancy\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/max> <http://www.w3.org/ns/mls#hasValue> \"988.7573262532552\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"13.404314994812012\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/count> <http://www.w3.org/ns/mls#hasValue> \"100\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/min>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"645.8515459872854\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/count>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/min>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2/std>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/std>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/mean>\n",
      "<http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a> <http://purl.org/dc/terms/identifier> \"model/model_24e6a2b6b9c64aec9605bc00b2ed3b5a.h5\"\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_cac90a7b1bf147bda94625edc5b339f7>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#CO2_Level>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/std>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/max> <http://www.w3.org/ns/mls#hasValue> \"992.8106770833333\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/std> <http://www.w3.org/ns/mls#hasValue> \"95.65189043806267\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#model_0754017c705548d490065f71ac4364ee>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/count>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"8.484709955579095\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TestDataset>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/count> <http://www.w3.org/ns/mls#hasValue> \"692\"^^<http://www.w3.org/2001/XMLSchema#long>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/mean>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/std>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://brickschema.org/schema/1.0.3/Brick#Occupancy>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"552.1107061382434\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/std>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#executes> <http://inter-space.org/learning_model#tensorflow>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2> <http://www.w3.org/2000/01/rdf-schema#label> \"Room CO2 Level\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"6.6768574714660645\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/min>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy> <http://www.w3.org/2000/01/rdf-schema#label> \"Room Occupancy\"^^<http://www.w3.org/2001/XMLSchema#string>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/max> <http://www.w3.org/ns/mls#hasValue> \"1122.069374593099\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"5.497774124145508\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/occupancy>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/std>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#achieves> <http://inter-space.org/learning_model#occupancy_task>\n",
      "<http://inter-space.org/learning_model#run_0d089607b4a04af1b08bd5693ecfac80> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#model_24e6a2b6b9c64aec9605bc00b2ed3b5a>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/max>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval/mean_squared_error> <http://www.w3.org/ns/mls#hasValue> \"129.43197631835938\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"640.6467203369142\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/std>\n",
      "<http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_26785a849ddd491ab2dded4bdaf7bb6b/eval>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/count>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/std> <http://www.w3.org/ns/mls#hasValue> \"19.768259719433466\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"551.0129083150229\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/co2> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Feature>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/max>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"78.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e> <http://www.w3.org/ns/mls#hasSplit> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#realizes> \"tensorflow.python.keras.engine.sequential.Sequential\"\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#run_db32906e6fa04a8db3fa43cf57d7f447/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"29.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval/mean_absolute_error> <http://www.w3.org/ns/mls#hasValue> \"9.758745193481445\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2/mean> <http://www.w3.org/ns/mls#hasValue> \"639.8968543802897\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/occupancy> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/occ_sensors/3>\n",
      "<http://inter-space.org/learning_model#run_57076ea6d4f54a0d9ae368b28b083d2a/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TestDataset>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/count>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_ec01946ecb4f4ae58d1397bb39eb20d2/eval>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/mean> <http://www.w3.org/ns/mls#hasValue> \"8.8\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/occupancy/std>\n",
      "<http://inter-space.org/learning_model#run_ee12a1f76fd14558ab791db83b180979/eval> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#ModelEvaluation>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"26.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/test/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/mean> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanValue>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_squared_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/count>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#Column>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#TrainDataset>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/test/co2/mean>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/mean>\n",
      "<http://inter-space.org/learning_model#run_0cc6c3f70b0549a3aab1f9a8092af586> <http://www.w3.org/ns/mls#hasInput> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/occupancy/std> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#StdValue>\n",
      "<http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_c1f763a8e44a42e4aae82464640e80b5/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/occupancy> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/occupancy>\n",
      "<http://inter-space.org/learning_model#run_0d044d9ef34440598ad5a17e04d92dbe/eval/mean_absolute_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanAbsoluteError>\n",
      "<http://inter-space.org/learning_model#run_a804900960cd4f1a852eb8d94b47343c/eval/mean_squared_error> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MeanSquaredError>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/co2/max> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MaxValue>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/finetune/occupancy/max> <http://www.w3.org/ns/mls#hasValue> \"33.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#run_5561bfe9fe5c4d78979a8a28882f356f/eval/mean_absolute_error>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/min> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#MinValue>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/min>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/co2> <https://brickschema.org/schema/1.0.3/BrickFrame#isMeasuredBy> <http://inter-space.org/learning_model#/co2_sensors/3>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/co2/max>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/train/co2/std>\n",
      "<http://inter-space.org/learning_model#fa639abdd2e940149afd422aab51e5d8/train/co2/min> <http://www.w3.org/ns/mls#hasValue> \"410.78400065104165\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862> <http://www.w3.org/ns/mls#hasOutput> <http://inter-space.org/learning_model#run_205f14470085424e830cd1c5b7cc4862/eval>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2/min>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy> <http://www.w3.org/ns/mls#hasQuality> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/test/occupancy/std>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/train/occupancy/min> <http://www.w3.org/ns/mls#hasValue> \"0.0\"^^<http://www.w3.org/2001/XMLSchema#float>\n",
      "<http://inter-space.org/learning_model#49117df88973460f947218e5f0adb18e/finetune/count> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/mls#DatasetCharacteristic>\n",
      "<http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/finetune/co2> <http://www.w3.org/ns/mls#specifiedBy> <http://inter-space.org/learning_model#b782a3a359e54fcebbd57d590665bc39/co2>\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for line in lines:\n",
    "    elms = line.split(' ')\n",
    "    n = len(elms)\n",
    "    print(' '.join(elms[:n-2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.cbook as cbook\n",
    "\n",
    "# Load a numpy record array from yahoo csv data with fields date, open, close,\n",
    "# volume, adj_close from the mpl-data/example directory. The record array\n",
    "# stores the date as an np.datetime64 with a day unit ('D') in the date column.\n",
    "price_data = (cbook.get_sample_data('goog.npz', np_load=True)['price_data']\n",
    "              .view(np.recarray))\n",
    "price_data = price_data[-250:]  # get the most recent 250 trading days\n",
    "\n",
    "delta1 = np.diff(price_data.adj_close) / price_data.adj_close[:-1]\n",
    "\n",
    "# Marker size in units of points^2\n",
    "volume = (15 * price_data.volume[:-2] / price_data.volume[0])**2\n",
    "close = 0.003 * price_data.close[:-2] / 0.003 * price_data.open[:-2]\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.scatter(delta1[:-1], delta1[1:], c=close, s=volume, alpha=0.5)\n",
    "\n",
    "ax.set_xlabel(r'$\\Delta_i$', fontsize=15)\n",
    "ax.set_ylabel(r'$\\Delta_{i+1}$', fontsize=15)\n",
    "ax.set_title('Volume and percent change')\n",
    "\n",
    "ax.grid(True)\n",
    "fig.tight_layout()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 487,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Models')"
      ]
     },
     "execution_count": 487,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x640 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8, 8), dpi=80)\n",
    "plt.scatter(values, models)\n",
    "plt.xlabel(metric)\n",
    "plt.ylabel('Models')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 485,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "---------------\n",
      "[167.99900817871094, 9.758745193481445, 167.99900817871094]\n",
      "mean_absolute_error 1 9.758745193481445\n",
      "[63.64848709106445, 5.497774124145508, 63.64848709106445]\n",
      "mean_absolute_error 2 5.497774124145508\n",
      "[129.43197631835938, 9.648994445800781, 129.43197631835938]\n",
      "mean_absolute_error 3 9.648994445800781\n",
      "---------------\n",
      "[327.7085876464844, 13.404314994812012, 327.7085876464844]\n",
      "mean_absolute_error 1 13.404314994812012\n",
      "[21.834674835205078, 3.7022125720977783, 21.834674835205078]\n",
      "mean_absolute_error 2 3.7022125720977783\n",
      "[65.63717651367188, 6.6768574714660645, 65.63717651367188]\n",
      "mean_absolute_error 3 6.6768574714660645\n",
      "---------------\n",
      "[328.2415771484375, 13.324183464050293, 328.2415771484375]\n",
      "mean_absolute_error 1 13.324183464050293\n",
      "[34.17942810058594, 4.665645599365234, 34.17942810058594]\n",
      "mean_absolute_error 2 4.665645599365234\n",
      "[40.540653228759766, 5.017570495605469, 40.540653228759766]\n",
      "mean_absolute_error 3 5.017570495605469\n"
     ]
    }
   ],
   "source": [
    "#room_id = 1\n",
    "models = []\n",
    "values = []\n",
    "#metric = 'mean_squared_error'\n",
    "metric = 'mean_absolute_error'\n",
    "for source_room in rooms:\n",
    "    \n",
    "    \n",
    "    source_model = rooms[source_room]['model']\n",
    "    \n",
    "    \n",
    "    #source_model_id = rooms[source_room]['model_uuid']\n",
    "    #source_model_filename = f'model/model_{source_model_id}.h5'\n",
    "    #source_model = tf.keras.models.load_model(source_model_filename)    \n",
    "\n",
    "    print('---------------')\n",
    "    # add test evaluations\n",
    "    for target_room in rooms:\n",
    "        \n",
    "        metric_values = source_model.evaluate(rooms[target_room]['test_data'], rooms[target_room]['test_labels'], verbose=0)\n",
    "        print(metric_values)\n",
    "        m_ind = 0\n",
    "        for m in source_model.metrics:\n",
    "            if m.name == metric:\n",
    "                models.append(f'Model {source_room}')\n",
    "                values.append(metric_values[m_ind])\n",
    "                print(m.name, target_room, metric_values[m_ind])\n",
    "            m_ind += 1\n",
    "            \n",
    "        \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 463,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[9.758745193481445,\n",
       " 5.497774124145508,\n",
       " 9.648994445800781,\n",
       " 13.404314994812012,\n",
       " 3.7022125720977783,\n",
       " 6.6768574714660645,\n",
       " 13.324183464050293,\n",
       " 4.665645599365234,\n",
       " 5.017570495605469]"
      ]
     },
     "execution_count": 463,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "rooms = {\n",
    "    1: {'dataset_uuid':  'b782a3a359e54fcebbd57d590665bc39' , 'model_uuid': '0754017c705548d490065f71ac4364ee' },\n",
    "    2: {'dataset_uuid':  '49117df88973460f947218e5f0adb18e' , 'model_uuid': '24e6a2b6b9c64aec9605bc00b2ed3b5a' },\n",
    "    3: {'dataset_uuid':  'fa639abdd2e940149afd422aab51e5d8' , 'model_uuid': 'cac90a7b1bf147bda94625edc5b339f7' }\n",
    "}\n",
    "\n",
    "\n",
    "for room_id in rooms:\n",
    "    \n",
    "    # load model\n",
    "    model_uuid = rooms[room_id]['model_uuid']\n",
    "    model_file = f'model/model_{model_uuid}.h5'\n",
    "    rooms[room_id]['model_file'] = model_file\n",
    "    rooms[room_id]['model'] = tf.keras.models.load_model(model_file)\n",
    "    \n",
    "    # load dataset\n",
    "    dataset_uuid = rooms[room_id]['dataset_uuid']\n",
    "    dataset_filename = f'dataset/ds_{dataset_uuid}.h5'\n",
    "    ds = pd.HDFStore(dataset_filename, 'r')\n",
    "\n",
    "    test_dataset = ds['test']\n",
    "    rooms[room_id]['test_data'] = test_dataset\n",
    "    rooms[room_id]['test_labels'] = test_dataset.pop('occupancy')\n",
    "\n",
    "    finetune_dataset = ds['finetune']\n",
    "    rooms[room_id]['finetune_data'] = finetune_dataset\n",
    "    rooms[room_id]['finetune_labels'] = finetune_dataset.pop('occupancy')\n",
    "    \n",
    "    \n",
    "    \n",
    "    ds.close()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont = Ontology()\n",
    "ont.load_ttl('ontology/base_graph.ttl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 167.9990 - mean_absolute_error: 9.7587 - mean_squared_error: 167.9990\n",
      "4/4 - 0s - loss: 63.6485 - mean_absolute_error: 5.4978 - mean_squared_error: 63.6485\n",
      "4/4 - 0s - loss: 129.4320 - mean_absolute_error: 9.6490 - mean_squared_error: 129.4320\n",
      "4/4 - 0s - loss: 327.7086 - mean_absolute_error: 13.4043 - mean_squared_error: 327.7086\n",
      "4/4 - 0s - loss: 21.8347 - mean_absolute_error: 3.7022 - mean_squared_error: 21.8347\n",
      "4/4 - 0s - loss: 65.6372 - mean_absolute_error: 6.6769 - mean_squared_error: 65.6372\n",
      "4/4 - 0s - loss: 328.2416 - mean_absolute_error: 13.3242 - mean_squared_error: 328.2416\n",
      "4/4 - 0s - loss: 34.1794 - mean_absolute_error: 4.6656 - mean_squared_error: 34.1794\n",
      "4/4 - 0s - loss: 40.5407 - mean_absolute_error: 5.0176 - mean_squared_error: 40.5407\n"
     ]
    }
   ],
   "source": [
    "# add dataset and train runs\n",
    "for room_id, room in rooms.items():\n",
    "    \n",
    "    brick_ref = {\n",
    "        'co2': [\n",
    "            ('rdf:type', 'brick:CO2_Level'), \n",
    "            ('rdfs:label', 'Room CO2 Level'),\n",
    "            ('brickframe:isMeasuredBy', 'isl:/co2_sensors/'+str(room_id))\n",
    "        ],\n",
    "        'occupancy': [\n",
    "            ('rdf:type', 'brick:Occupancy'),\n",
    "            ('rdfs:label', 'Room Occupancy'),\n",
    "            ('brickframe:isMeasuredBy', 'isl:/occ_sensors/'+str(room_id))\n",
    "        ] \n",
    "    }    \n",
    "    \n",
    "\n",
    "    ont.add_dataset(room['dataset_uuid'], brick_ref)\n",
    "    \n",
    "    \n",
    "    # add train run\n",
    "    model_uuid = room['model_uuid']\n",
    "    \n",
    "    train_split_id = room['dataset_uuid'] + '/train'\n",
    "    ont.add_run(room['model_file'], 'occupancy_task', train_split_id)\n",
    "\n",
    "    \n",
    "# add test runs (source room model + target room data)\n",
    "for source_room in rooms:\n",
    "    \n",
    "    source_model = rooms[source_room]['model']\n",
    "\n",
    "    for target_room in rooms:\n",
    "        \n",
    "        metric_values = source_model.evaluate(rooms[target_room]['test_data'], rooms[target_room]['test_labels'], verbose=2)\n",
    "        metric_names = [m.name for m in source_model.metrics]\n",
    "        metrics = dict(zip(metric_names, metric_values))\n",
    "\n",
    "        test_split_id = rooms[target_room]['dataset_uuid'] + '/test'\n",
    "        ont.add_run(rooms[source_room]['model_file'], 'occupancy_task', test_split_id, metrics)    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 - 0s - loss: 52.9462 - mean_absolute_error: 5.6570 - mean_squared_error: 52.9462\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "rdflib.term.URIRef('http://inter-space.org/learning_model#run_9a03176913264f83bc8bdd6485fba690')"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# add fine-tune training run (room 2 model fine-tuned by room 3 data)\n",
    "model_room_id = 2\n",
    "ds_room_id = 3\n",
    "\n",
    "orig_model_id = rooms[model_room_id]['model_uuid']\n",
    "#new_model_file = f'model/model_{orig_model_id}.h5'\n",
    "#new_model = tf.keras.models.load_model(new_model_file)\n",
    "\n",
    "finetuned_model_id = 'b4895793546a4048b0656859e1430349'\n",
    "finetuned_model_file = f'model/model_{finetuned_model_id}.h5'\n",
    "finetuned_model = tf.keras.models.load_model(finetuned_model_file)\n",
    "\n",
    "train_split_id = rooms[ds_room_id]['dataset_uuid'] + '/finetune'\n",
    "test_split_id = rooms[ds_room_id]['dataset_uuid'] + '/test'\n",
    "\n",
    "# add training run\n",
    "ont.add_run(finetuned_model_file, 'occupancy_task', train_split_id, built_on = orig_model_id)\n",
    "\n",
    "# evaluate new fine-tuned model for room 3 test data\n",
    "metric_values = finetuned_model.evaluate(rooms[ds_room_id]['test_data'], rooms[ds_room_id]['test_labels'], verbose=2)\n",
    "metric_names = [m.name for m in finetuned_model.metrics]\n",
    "metrics = dict(zip(metric_names, metric_values))\n",
    "\n",
    "# add test evaluation of fine-tuned model to ontology\n",
    "ont.add_run(finetuned_model_file, 'occupancy_task', test_split_id, metrics) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.nq')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "ont.save('building_model_1.ttl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'model_0754017c705548d490065f71ac4364ee': {'dataset': 'b782a3a359e54fcebbd57d590665bc39', 'built_on': None}, 'model_24e6a2b6b9c64aec9605bc00b2ed3b5a': {'dataset': '49117df88973460f947218e5f0adb18e', 'built_on': None}, 'model_cac90a7b1bf147bda94625edc5b339f7': {'dataset': 'fa639abdd2e940149afd422aab51e5d8', 'built_on': None}, 'model_b4895793546a4048b0656859e1430349': {'dataset': 'fa639abdd2e940149afd422aab51e5d8', 'built_on': '24e6a2b6b9c64aec9605bc00b2ed3b5a'}}\n"
     ]
    }
   ],
   "source": [
    "print(ont.models)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}