netdata-pandas-examples/anomalies_collector_deepdive.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Anomalies collector - deepdive tutorial!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://colab.research.google.com/github/netdata/community/blob/main/netdata-agent-api/netdata-pandas/anomalies_collector_deepdive.ipynb\" target=\"_blank\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This notebook will walk through, step by step, a worked example of how the [netdata anomalies collector](https://github.com/netdata/netdata/tree/master/collectors/python.d.plugin/anomalies) works under the hood. \n",
    "\n",
    "**Note**: you can click the \"Open in Colab\" button above to open this notebook in [Google Colab](https://colab.research.google.com/notebooks/intro.ipynb#recent=true) where you can just get going with it without having to set up python environments or any messy stuff like that. If the button does not work then just [click here](https://colab.research.google.com/github/netdata/community/blob/main/netdata-agent-api/netdata-pandas/anomalies_collector_deepdive.ipynb).\n",
    "\n",
    "**Another Note**: If you are just reading through this notebook then it might be better to view it using nbviewer [here](https://nbviewer.jupyter.org/github/netdata/community/blob/main/netdata-agent-api/netdata-pandas/anomalies_collector_deepdive.ipynb) or colab [here](https://colab.research.google.com/github/netdata/community/blob/main/netdata-agent-api/netdata-pandas/anomalies_collector_deepdive.ipynb) as it will render a bit prettier than in Github."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# uncomment the below line to install required packages if needed (you will need to do this the first time if running in Google Colab).\n",
    "#!pip install netdata-pandas==0.0.28 numba==0.50.1 scikit-learn==0.23.2 pyod==0.8.3"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Contents:\n",
    "- [Overview](#Overview)\n",
    "- [Inputs & configuration](#Inputs-&-configuration)\n",
    "- [An aside on PCA](#An-aside-on-PCA)\n",
    "- [Initialize our models](#Initialize-our-models)\n",
    "- [Get training data](#Get-training-data)\n",
    "- [Preprocess or \"featurize\" the training data](#Preprocess-or-\"featurize\"-the-training-data)\n",
    "- [Train models](#Train-models)\n",
    "- [Get pediction data](#Get-pediction-data)\n",
    "- [Get predictions](#Get-predictions)\n",
    "- [But what _is_ the model?](#But-what-is-the-model?)\n",
    "- [Ok so lets step through that!](#Ok-so-lets-step-through-that!)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "But first, lets start with a meme..."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![image](img/mlfunny.jpg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Overview"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are three main concepts central to what the anomalies collector does:\n",
    "\n",
    "- **Featurization**: This is how we take the raw data for each chart and preprocess it into a feature representation or \"[feature vector](https://en.wikipedia.org/wiki/Feature_(machine_learning\\))\" used by the model. A simple way to think of this is that we just take each row of data and add some extra columns to encode some additional information. For example, a smoothed average of the last `lags_n` values for each dimension on the chart so the model can have some knowledge of the recent past beyond just the latest raw values of the dimensions.\n",
    "\n",
    "- **Training**: A function to take our \"featurized\" training data and train our models, one for each chart (or [custom models](https://github.com/netdata/netdata/tree/master/collectors/python.d.plugin/anomalies#custom-models) if you have defined any). This function will do slightly different things depending on what [model](https://pyod.readthedocs.io/en/latest/pyod.models.html#) you use. In a broad sense, its job is to train the model to form a useful, more compact, representation of the training data and then we can use this representation to measure our level of surprise at new data that we want to get anomaly scores for. So for the default PCA model this involves leveraging finding a lower dimensional representation that does a good job at reconstructing the main characteristics of the variance in our training data. Some other models might take a slightly different approach use different representations and algorithms to get to that \"measure of surprise\" for each feature vector. For the purpose of what we are doing this is largely abstracted away by the [API of the PyOD library](https://pyod.readthedocs.io/en/latest/pyod.html#api-reference), such that as a user we can easily switch between various models and still have broadly the same inputs ([Numpy arrays](https://numpy.org/doc/stable/reference/generated/numpy.array.html)) and outputs (also Numpy array's of anomaly scores, probabilities, and flags).\n",
    "\n",
    "- **Prediction**: Each trained model has a [`predict()`](https://pyod.readthedocs.io/en/latest/api_cc.html#pyod.models.base.BaseDetector.predict) function that we can use by passing in a new feature vector and getting back an anomaly probability and anomaly flag from the trained model. This is the part where we actually use the trained model as new data arrives to ask - \"how unusual does this new data look?\"  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Lets go!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "from datetime import datetime\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from IPython.display import display, Markdown\n",
    "from IPython.lib.display import YouTubeVideo\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from netdata_pandas.data import get_data, get_allmetrics\n",
    "from pyod.models.hbos import HBOS\n",
    "from pyod.models.pca import PCA\n",
    "from pyod.models.cblof import CBLOF\n",
    "from pyod.models.iforest import IForest\n",
    "\n",
    "\n",
    "def make_features(df, lags_n, diffs_n, smooth_n):\n",
    "    \"\"\"Given a pandas dataframe preprocess it to take differences, add smoothing, and lags as specified. \n",
    "    \"\"\"\n",
    "    if diffs_n >= 1:\n",
    "        # take differences\n",
    "        df = df.diff(diffs_n).dropna()\n",
    "    if smooth_n >= 2:\n",
    "        # apply a rolling average to smooth out the data a bit\n",
    "        df = df.rolling(smooth_n).mean().dropna()\n",
    "    if lags_n >= 1:\n",
    "        # for each dimension add a new columns for each of lags_n lags of the differenced and smoothed values for that dimension\n",
    "        df_columns_new = [f'{col}_lag{n}' for n in range(lags_n+1) for col in df.columns]\n",
    "        df = pd.concat([df.shift(n) for n in range(lags_n + 1)], axis=1).dropna()\n",
    "        df.columns = df_columns_new\n",
    "    # sort columns to have lagged values next to each other for clarity when looking at the feature vectors\n",
    "    df = df.reindex(sorted(df.columns), axis=1)\n",
    "    \n",
    "    return df\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inputs & configuration"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the next cell we will define all the inputs we will use in this tutorial. Feel free to play with them once you are familiar with how it all hangs together.\n",
    "\n",
    "Below you will see that the paramater values map to a subset of the inputs (the most important ones to help explain whats going on) required as part of the [`anomalies.conf`](https://github.com/netdata/netdata/tree/master/collectors/python.d.plugin/anomalies/anomalies.conf) configuration for the [anomalies collector](https://github.com/netdata/netdata/tree/master/collectors/python.d.plugin/anomalies) itself."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# inputs\n",
    "\n",
    "# what host will we use\n",
    "host = 'london.my-netdata.io'\n",
    "#host= 'cdn77.my-netdata.io'\n",
    "#host = 'odyslam.me/netdata/webserver'\n",
    "\n",
    "# for this tutorial we will just use a few charts\n",
    "charts_in_scope = ['system.cpu', 'system.load', 'system.net', 'system.io', 'system.ip', 'system.idlejitter']\n",
    "\n",
    "# what model from PyOD will we use under the hood\n",
    "model = 'pca'\n",
    "\n",
    "# how many seconds of data will we train our models on\n",
    "train_n_secs = 14400\n",
    "\n",
    "# what contamination rate will we use, see some discussion here to understand this one more: https://github.com/yzhao062/pyod/issues/144\n",
    "contamination = 0.001\n",
    "\n",
    "# if we want to ignore a recent window of data when training the model we can use this\n",
    "offset_n_secs = 0\n",
    "\n",
    "# how many lags to include in our feature vector\n",
    "lags_n = 5\n",
    "\n",
    "# how much smoothing to apply in our feature vector\n",
    "smooth_n = 3\n",
    "\n",
    "# if we want to do everything in terms of differences then we set diffs_n=1\n",
    "diffs_n = 1\n",
    "\n",
    "# for purpose of this turorial how many prediction steps will we take once we have a trained model\n",
    "n_prediction_steps = 20"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## An aside on PCA"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "By default the anomalies collector uses the `PCA` model, primarily this is because the PCA model gives a good combination of being able to capture and model flexible patterns in the data while also being computationally cheap since under the hood it is using the well researched, optimized and understood [SVD](https://en.wikipedia.org/wiki/Singular_value_decomposition) algorithm to decompose our featurized data and project it onto a lower dimensional space. At a high level, when we see new data that is in a strange or unexpected part of this lower dimensional space then this is symptomatic of some anomalous data and so will get a higher anomaly score. \n",
    "\n",
    "- api: https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.pca\n",
    "- source: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/pca.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: If you want to learn more about PCA and play with some notebooks exploring PCA in a similar manner to this one then check out [this chapter](https://jakevdp.github.io/PythonDataScienceHandbook/05.09-principal-component-analysis.html) from the great [Python Data Science Handbook](https://jakevdp.github.io/PythonDataScienceHandbook) by [Jake Vander Plas](https://twitter.com/jakevdp).\n",
    "\n",
    "The image below (taken from the book) gives a good intuition about a way of thinking of PCA as (almost) dimensionality reduction. In the image below we are looking at how PCA could be used to \"compress\" the X and Y data into one single dimension of numbers by projecting each pair of points onto the corresponding solid blue line of dots. \n",
    "\n",
    "![image](img/pca1.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Or if a [Computerfile](https://www.youtube.com/channel/UC9-y-6csu5WGm29I7JiwpnA) video is more your thing then check out the below one as it does a better job then we can here."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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\n",
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"400\"\n",
       "            height=\"300\"\n",
       "            src=\"https://www.youtube.com/embed/TJdH6rPA-TI\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.YouTubeVideo at 0x1ba9c2846a0>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "YouTubeVideo('TJdH6rPA-TI')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this sense the PCA is learning a compressed representation of the data (technically, as mentioned in the video above, its a transformed representation in a different space, but we dont need to get too bogged down in that for our purposes here). This is essentially how PyOD uses PCA for anomaly detection under the hood. \n",
    "\n",
    "So we use PCA to learn a compressed representation of all the training data for each chart is some more abstract lower dimensional space that can be of use to us. The \"of use to us\" part here comes from the fact that PCA picks these lower dimensional representations in a clever way (using [SVD](https://en.wikipedia.org/wiki/Singular_value_decomposition)) that maximises the amount of information about the main directions of variance of the original dataset. e.g. in the image above the line of blue dots is the best line we could project our data onto in order to maintain as much information as possible about the variance of the original 2 dimensional dataset, as you can see there is much more 'spread' in the data along the original x-axis than the y-axis and that is what is captured by the 'spread' in the solid blue line of dots.  \n",
    "\n",
    "So when we see a new observation if it does not map well into this lower level representation we have learned during training, then that tells us that this new observation does not fit well into the representation we learned during training and as such it's probably somewhat anomalous, at least in comparison to what we observed in general in our training data. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: We have looked at PCA a little bit above, but the point of awesome libraries like PyOD is that you don't really need to go that far down into the details - once you understand how the API works and a little bit about the various types of models/approaches used you can consider playing with and trying out other models on your data that take completely different approaches under the hood, for example:\n",
    "\n",
    "- **hbos**: uses histograms as the underlying representations of your data used to then measure the surprise of new data ([more info](https://www.dfki.de/fileadmin/user_upload/import/6431_HBOS-poster.pdf)).\n",
    "- **cblof**: somewhat similar in approach to pca but learned clusters as the representation of your training data and then it's the distance of new observations to the learned cluster centroids that is used to generate an anomaly score ([a good blog post](http://towardsdatascience.com/local-outlier-factor-for-anomaly-detection-cc0c770d2ebe)).\n",
    "- **iforest**: uses a [isolation forest](https://en.wikipedia.org/wiki/Isolation_forest) as the underlying model and then observations that end up on strnage partso f that learned forest are then considered more anomalous ([sklearn user guide](https://scikit-learn.org/stable/modules/outlier_detection.html#isolation-forest)). \n",
    "- ...[etc](https://pyod.readthedocs.io/en/latest/pyod.models.html).\n",
    "\n",
    "**Note**: Not all models from PyOD have been implemented in the anomalies collector as some turned out to be too expensive for the specific use case of unsupervised anomaly detection on your Netdata node itself (or even on a parent node). To that end, the models available in the collector are [`pca`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.pca), [`hbos`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.hbos), [`iforest`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.iforest), [`cblof`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.cblof), [`loda`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.loda), [`copod`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.copod) or [`feature_bagging`](https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.feature_bagging).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Initialize our models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will initialize a PyOD model for each chart in `charts_in_scope`. Each model type in PyOD will have various different input parameters that a user can play with, we will tend to use the defaults and override them sometimes with ones we have picked based on what we know about the task we are working on. Generally (where relevant) these model parameters, apart from `contamination`, are hardcoded into the anomalies collector based on our internal research as we developed the collector, you can see this in the [collector code here](https://github.com/netdata/netdata/blob/master/collectors/python.d.plugin/anomalies/anomalies.chart.py#L133).\n",
    "\n",
    "In the cell below we have added a comment for the source and API reference of each model from PyOD so you can take a look and read more about each one."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initialize a model for each chart\n",
    "if model == 'pca':\n",
    "    # api: https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.pca\n",
    "    # source: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/pca.html\n",
    "    models = {c: PCA(contamination=contamination, n_components=2, n_selected_components=2) for c in charts_in_scope}\n",
    "elif model == 'hbos':\n",
    "    # api: https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.hbos\n",
    "    # source: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/hbos.html\n",
    "    models = {c: HBOS(contamination=contamination) for c in charts_in_scope}\n",
    "elif model == 'cblof':\n",
    "    # api: https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.cblof\n",
    "    # source: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/cblof.html\n",
    "    models = {c: CBLOF(contamination=contamination, n_clusters=4) for c in charts_in_scope}\n",
    "elif model == 'iforest':\n",
    "    # api: https://pyod.readthedocs.io/en/latest/pyod.models.html#module-pyod.models.iforest\n",
    "    # source: https://pyod.readthedocs.io/en/latest/_modules/pyod/models/iforest.html\n",
    "    models = {c: IForest(contamination=contamination, n_estimators=50, bootstrap=True, behaviour='new') for c in charts_in_scope}\n",
    "else:\n",
    "    # we used the HBOS as default as it is both fast and robust to many different types of data and has proven in internal development \n",
    "    # to have less failure modes then some other models given the wide variaty of data we are expecting to be thrown at it\n",
    "    models = {c: HBOS(contamination=contamination) for c in charts_in_scope}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get training data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first thing we need to do is get our raw training data for each chart we want to build a model for.\n",
    "\n",
    "To get the data we will make use of the [netdata-pandas](https://github.com/netdata/netdata-pandas) library we have built to make multiple asynchronous calls (using [asks](https://github.com/theelous3/asks) and [trio](https://github.com/python-trio/trio)) to the [Netdata REST API](https://learn.netdata.cloud/docs/agent/web/api) and basically wrangle the results into a nice [Pandas](https://pandas.pydata.org/) [`DataFrame`](https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.html)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 14399 entries, 1604474576 to 1604488974\n",
      "Data columns (total 21 columns):\n",
      " #   Column                     Non-Null Count  Dtype  \n",
      "---  ------                     --------------  -----  \n",
      " 0   system.cpu|guest           14399 non-null  float32\n",
      " 1   system.cpu|guest_nice      14399 non-null  float32\n",
      " 2   system.cpu|iowait          14399 non-null  float32\n",
      " 3   system.cpu|irq             14399 non-null  float32\n",
      " 4   system.cpu|nice            14399 non-null  float32\n",
      " 5   system.cpu|softirq         14399 non-null  float32\n",
      " 6   system.cpu|steal           14399 non-null  float32\n",
      " 7   system.cpu|system          14399 non-null  float32\n",
      " 8   system.cpu|user            14399 non-null  float32\n",
      " 9   system.idlejitter|average  14399 non-null  float32\n",
      " 10  system.idlejitter|max      14399 non-null  float32\n",
      " 11  system.idlejitter|min      14399 non-null  float32\n",
      " 12  system.io|in               14399 non-null  float32\n",
      " 13  system.io|out              14399 non-null  float32\n",
      " 14  system.ip|received         14399 non-null  float32\n",
      " 15  system.ip|sent             14399 non-null  float32\n",
      " 16  system.load|load1          14395 non-null  float32\n",
      " 17  system.load|load15         14395 non-null  float32\n",
      " 18  system.load|load5          14395 non-null  float32\n",
      " 19  system.net|received        14399 non-null  float32\n",
      " 20  system.net|sent            14399 non-null  float32\n",
      "dtypes: float32(21)\n",
      "memory usage: 1.3 MB\n",
      "None\n"
     ]
    },
    {
     "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>system.cpu|guest</th>\n",
       "      <th>system.cpu|guest_nice</th>\n",
       "      <th>system.cpu|iowait</th>\n",
       "      <th>system.cpu|irq</th>\n",
       "      <th>system.cpu|nice</th>\n",
       "      <th>system.cpu|softirq</th>\n",
       "      <th>system.cpu|steal</th>\n",
       "      <th>system.cpu|system</th>\n",
       "      <th>system.cpu|user</th>\n",
       "      <th>system.idlejitter|average</th>\n",
       "      <th>...</th>\n",
       "      <th>system.idlejitter|min</th>\n",
       "      <th>system.io|in</th>\n",
       "      <th>system.io|out</th>\n",
       "      <th>system.ip|received</th>\n",
       "      <th>system.ip|sent</th>\n",
       "      <th>system.load|load1</th>\n",
       "      <th>system.load|load15</th>\n",
       "      <th>system.load|load5</th>\n",
       "      <th>system.net|received</th>\n",
       "      <th>system.net|sent</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604474576</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>82.0</td>\n",
       "      <td>...</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-11.750492</td>\n",
       "      <td>110.332367</td>\n",
       "      <td>-174.585098</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20.880480</td>\n",
       "      <td>-84.349640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474577</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.505050</td>\n",
       "      <td>0.505050</td>\n",
       "      <td>84.0</td>\n",
       "      <td>...</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>192.250900</td>\n",
       "      <td>-64.668961</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>164.099686</td>\n",
       "      <td>-32.041279</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474578</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.251256</td>\n",
       "      <td>1.005025</td>\n",
       "      <td>78.0</td>\n",
       "      <td>...</td>\n",
       "      <td>27.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>170.466507</td>\n",
       "      <td>-86.694138</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>124.846687</td>\n",
       "      <td>-37.602589</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474579</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.502513</td>\n",
       "      <td>0.502513</td>\n",
       "      <td>79.0</td>\n",
       "      <td>...</td>\n",
       "      <td>29.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>153.144745</td>\n",
       "      <td>-199.074005</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>46.703060</td>\n",
       "      <td>-91.418747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474580</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.503778</td>\n",
       "      <td>0.251889</td>\n",
       "      <td>76.0</td>\n",
       "      <td>...</td>\n",
       "      <td>47.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-26.941219</td>\n",
       "      <td>124.466759</td>\n",
       "      <td>-144.531433</td>\n",
       "      <td>0.03</td>\n",
       "      <td>0.02</td>\n",
       "      <td>0.07</td>\n",
       "      <td>39.373100</td>\n",
       "      <td>-58.933270</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|guest  system.cpu|guest_nice  system.cpu|iowait  \\\n",
       "time_idx                                                                 \n",
       "1604474576               0.0                    0.0                0.0   \n",
       "1604474577               0.0                    0.0                0.0   \n",
       "1604474578               0.0                    0.0                0.0   \n",
       "1604474579               0.0                    0.0                0.0   \n",
       "1604474580               0.0                    0.0                0.0   \n",
       "\n",
       "            system.cpu|irq  system.cpu|nice  system.cpu|softirq  \\\n",
       "time_idx                                                          \n",
       "1604474576             0.0              0.0                 0.0   \n",
       "1604474577             0.0              0.0                 0.0   \n",
       "1604474578             0.0              0.0                 0.0   \n",
       "1604474579             0.0              0.0                 0.0   \n",
       "1604474580             0.0              0.0                 0.0   \n",
       "\n",
       "            system.cpu|steal  system.cpu|system  system.cpu|user  \\\n",
       "time_idx                                                           \n",
       "1604474576               0.0           0.500000         0.500000   \n",
       "1604474577               0.0           0.505050         0.505050   \n",
       "1604474578               0.0           0.251256         1.005025   \n",
       "1604474579               0.0           0.502513         0.502513   \n",
       "1604474580               0.0           0.503778         0.251889   \n",
       "\n",
       "            system.idlejitter|average  ...  system.idlejitter|min  \\\n",
       "time_idx                               ...                          \n",
       "1604474576                       82.0  ...                   46.0   \n",
       "1604474577                       84.0  ...                   30.0   \n",
       "1604474578                       78.0  ...                   27.0   \n",
       "1604474579                       79.0  ...                   29.0   \n",
       "1604474580                       76.0  ...                   47.0   \n",
       "\n",
       "            system.io|in  system.io|out  system.ip|received  system.ip|sent  \\\n",
       "time_idx                                                                      \n",
       "1604474576           0.0     -11.750492          110.332367     -174.585098   \n",
       "1604474577           0.0       0.000000          192.250900      -64.668961   \n",
       "1604474578           0.0       0.000000          170.466507      -86.694138   \n",
       "1604474579           0.0       0.000000          153.144745     -199.074005   \n",
       "1604474580           0.0     -26.941219          124.466759     -144.531433   \n",
       "\n",
       "            system.load|load1  system.load|load15  system.load|load5  \\\n",
       "time_idx                                                               \n",
       "1604474576                NaN                 NaN                NaN   \n",
       "1604474577                NaN                 NaN                NaN   \n",
       "1604474578                NaN                 NaN                NaN   \n",
       "1604474579                NaN                 NaN                NaN   \n",
       "1604474580               0.03                0.02               0.07   \n",
       "\n",
       "            system.net|received  system.net|sent  \n",
       "time_idx                                          \n",
       "1604474576            20.880480       -84.349640  \n",
       "1604474577           164.099686       -32.041279  \n",
       "1604474578           124.846687       -37.602589  \n",
       "1604474579            46.703060       -91.418747  \n",
       "1604474580            39.373100       -58.933270  \n",
       "\n",
       "[5 rows x 21 columns]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# define the window for the training data to pull\n",
    "before = int(datetime.now().timestamp()) - offset_n_secs\n",
    "after =  before - train_n_secs\n",
    "\n",
    "# get the training data\n",
    "df_train = get_data(hosts=host, charts=charts_in_scope, after=after, before=before, sort_cols=True, numeric_only=True, float_size='float32')\n",
    "print(df_train.info())\n",
    "df_train.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Above we can see our raw training data is just a pandas `DataFrame` with a timestamp index and a column of floats for each dimension from our `charts_in_scope` list.\n",
    "\n",
    "**Note**: The [netdata-pandas](https://github.com/netdata/netdata-pandas) default naming convention for columns is \"chart.name|dimension.name\" "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocess or \"featurize\" the training data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Before we train our model we will first do some preprocessing to the raw data to create a \"feature vector\" to try and encode a more flexible and powerful representation for the model to work with as opposed to just looking at the most recently observed values in isolation. \n",
    "\n",
    "This is the \"featurization\" we mentioned at the begining of the notebook. The idea here is to give the model some extra information so that it may spot more complex and interesting anomalies as opposed to just spikes where one metric is a very high or very low value.   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(14387, 126)\n"
     ]
    },
    {
     "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>system.cpu|guest_lag0</th>\n",
       "      <th>system.cpu|guest_lag1</th>\n",
       "      <th>system.cpu|guest_lag2</th>\n",
       "      <th>system.cpu|guest_lag3</th>\n",
       "      <th>system.cpu|guest_lag4</th>\n",
       "      <th>system.cpu|guest_lag5</th>\n",
       "      <th>system.cpu|guest_nice_lag0</th>\n",
       "      <th>system.cpu|guest_nice_lag1</th>\n",
       "      <th>system.cpu|guest_nice_lag2</th>\n",
       "      <th>system.cpu|guest_nice_lag3</th>\n",
       "      <th>...</th>\n",
       "      <th>system.net|received_lag2</th>\n",
       "      <th>system.net|received_lag3</th>\n",
       "      <th>system.net|received_lag4</th>\n",
       "      <th>system.net|received_lag5</th>\n",
       "      <th>system.net|sent_lag0</th>\n",
       "      <th>system.net|sent_lag1</th>\n",
       "      <th>system.net|sent_lag2</th>\n",
       "      <th>system.net|sent_lag3</th>\n",
       "      <th>system.net|sent_lag4</th>\n",
       "      <th>system.net|sent_lag5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604474588</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-30.704399</td>\n",
       "      <td>-40.856167</td>\n",
       "      <td>6.443555</td>\n",
       "      <td>32.546180</td>\n",
       "      <td>32.694991</td>\n",
       "      <td>-12.822536</td>\n",
       "      <td>-24.871939</td>\n",
       "      <td>-14.788798</td>\n",
       "      <td>38.332296</td>\n",
       "      <td>7.027175</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474589</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>40.196271</td>\n",
       "      <td>-30.704399</td>\n",
       "      <td>-40.856167</td>\n",
       "      <td>6.443555</td>\n",
       "      <td>-7.484048</td>\n",
       "      <td>32.694991</td>\n",
       "      <td>-12.822536</td>\n",
       "      <td>-24.871939</td>\n",
       "      <td>-14.788798</td>\n",
       "      <td>38.332296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474590</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>37.268468</td>\n",
       "      <td>40.196271</td>\n",
       "      <td>-30.704399</td>\n",
       "      <td>-40.856167</td>\n",
       "      <td>-12.974803</td>\n",
       "      <td>-7.484048</td>\n",
       "      <td>32.694991</td>\n",
       "      <td>-12.822536</td>\n",
       "      <td>-24.871939</td>\n",
       "      <td>-14.788798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474591</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>2.695841</td>\n",
       "      <td>37.268468</td>\n",
       "      <td>40.196271</td>\n",
       "      <td>-30.704399</td>\n",
       "      <td>-20.702449</td>\n",
       "      <td>-12.974803</td>\n",
       "      <td>-7.484048</td>\n",
       "      <td>32.694991</td>\n",
       "      <td>-12.822536</td>\n",
       "      <td>-24.871939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474592</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-51.627127</td>\n",
       "      <td>2.695841</td>\n",
       "      <td>37.268468</td>\n",
       "      <td>40.196271</td>\n",
       "      <td>15.614039</td>\n",
       "      <td>-20.702449</td>\n",
       "      <td>-12.974803</td>\n",
       "      <td>-7.484048</td>\n",
       "      <td>32.694991</td>\n",
       "      <td>-12.822536</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 126 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|guest_lag0  system.cpu|guest_lag1  \\\n",
       "time_idx                                                   \n",
       "1604474588                    0.0                    0.0   \n",
       "1604474589                    0.0                    0.0   \n",
       "1604474590                    0.0                    0.0   \n",
       "1604474591                    0.0                    0.0   \n",
       "1604474592                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_lag2  system.cpu|guest_lag3  \\\n",
       "time_idx                                                   \n",
       "1604474588                    0.0                    0.0   \n",
       "1604474589                    0.0                    0.0   \n",
       "1604474590                    0.0                    0.0   \n",
       "1604474591                    0.0                    0.0   \n",
       "1604474592                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_lag4  system.cpu|guest_lag5  \\\n",
       "time_idx                                                   \n",
       "1604474588                    0.0                    0.0   \n",
       "1604474589                    0.0                    0.0   \n",
       "1604474590                    0.0                    0.0   \n",
       "1604474591                    0.0                    0.0   \n",
       "1604474592                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_nice_lag0  system.cpu|guest_nice_lag1  \\\n",
       "time_idx                                                             \n",
       "1604474588                         0.0                         0.0   \n",
       "1604474589                         0.0                         0.0   \n",
       "1604474590                         0.0                         0.0   \n",
       "1604474591                         0.0                         0.0   \n",
       "1604474592                         0.0                         0.0   \n",
       "\n",
       "            system.cpu|guest_nice_lag2  system.cpu|guest_nice_lag3  ...  \\\n",
       "time_idx                                                            ...   \n",
       "1604474588                         0.0                         0.0  ...   \n",
       "1604474589                         0.0                         0.0  ...   \n",
       "1604474590                         0.0                         0.0  ...   \n",
       "1604474591                         0.0                         0.0  ...   \n",
       "1604474592                         0.0                         0.0  ...   \n",
       "\n",
       "            system.net|received_lag2  system.net|received_lag3  \\\n",
       "time_idx                                                         \n",
       "1604474588                -30.704399                -40.856167   \n",
       "1604474589                 40.196271                -30.704399   \n",
       "1604474590                 37.268468                 40.196271   \n",
       "1604474591                  2.695841                 37.268468   \n",
       "1604474592                -51.627127                  2.695841   \n",
       "\n",
       "            system.net|received_lag4  system.net|received_lag5  \\\n",
       "time_idx                                                         \n",
       "1604474588                  6.443555                 32.546180   \n",
       "1604474589                -40.856167                  6.443555   \n",
       "1604474590                -30.704399                -40.856167   \n",
       "1604474591                 40.196271                -30.704399   \n",
       "1604474592                 37.268468                 40.196271   \n",
       "\n",
       "            system.net|sent_lag0  system.net|sent_lag1  system.net|sent_lag2  \\\n",
       "time_idx                                                                       \n",
       "1604474588             32.694991            -12.822536            -24.871939   \n",
       "1604474589             -7.484048             32.694991            -12.822536   \n",
       "1604474590            -12.974803             -7.484048             32.694991   \n",
       "1604474591            -20.702449            -12.974803             -7.484048   \n",
       "1604474592             15.614039            -20.702449            -12.974803   \n",
       "\n",
       "            system.net|sent_lag3  system.net|sent_lag4  system.net|sent_lag5  \n",
       "time_idx                                                                      \n",
       "1604474588            -14.788798             38.332296              7.027175  \n",
       "1604474589            -24.871939            -14.788798             38.332296  \n",
       "1604474590            -12.822536            -24.871939            -14.788798  \n",
       "1604474591             32.694991            -12.822536            -24.871939  \n",
       "1604474592             -7.484048             32.694991            -12.822536  \n",
       "\n",
       "[5 rows x 126 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# lets preprocess or \"featurize\" our raw data\n",
    "df_train_processed = make_features(df_train, lags_n, diffs_n, smooth_n)\n",
    "\n",
    "# print out the shape of our featurized data\n",
    "print(df_train_processed.shape)\n",
    "df_train_processed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The below few cells will explore a little what we have just done to try and make the ideas of preprocessing aka \"featurization\" aka \"feature vector\" a little clearer.\n",
    "\n",
    "Terms like \"featurization\" and \"feature vector\" are often used to sound fancy, but in reality its typically just as simple as adding additional columns to each row of your data where those new columns have numbers in them that represent something about your data that you want to make available to the model. \n",
    "\n",
    "So in our case adding lagged values of each smoothed and differenced dimension, is basically a design choice we make whereby we are telling the model we want it to consider `lags_n` recent values as opposed to just the latest observed dimensions. We do this because there are many [different types of anomalies](https://andrewm4894.com/2020/10/19/different-types-of-time-series-anomalies/) we want to try and be able to spot, so making a small snippet of recent data for each dimension available to the model gives us the ability to capture more complex anomaly patterns that might occur.\n",
    "\n",
    "If we were to just train the model on the most recent values for each dimension the best we could reasonably hope for it to capture would be anomalies where one or more dimension takes an unusually high or low value for one time step. This is essentially not that much better then a traditional approach using [z-scores](https://towardsdatascience.com/z-score-for-anomaly-detection-d98b0006f510). (If you are interested in comparing the two we actually also have a [zscores collector](https://github.com/andrewm4894/netdata/tree/zscores-collector/collectors/python.d.plugin/zscores) too, if, for example, you would like to just start simple or cannot install the ML Python libraries the anomalies collector depends on). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "df_train shape is (14399, 21)\n",
      "df_train_processed is (14387, 126)\n",
      "make_features has added 105 new columns, one for each lags_n (21*5=105)\n"
     ]
    }
   ],
   "source": [
    "# Lets look at how the shape of our data has changed due to preprocessing\n",
    "print(f'df_train shape is {df_train.shape}')\n",
    "print(f'df_train_processed is {df_train_processed.shape}')\n",
    "n_cols_added = len(df_train_processed.columns)-len(df_train.columns)\n",
    "print(f'make_features has added {n_cols_added} new columns, one for each lags_n ({df_train.shape[1]}*{lags_n}={n_cols_added})')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So as you can see from the above output, our featurization has added a new column for each `lags_n` specified. And we have also lost a few rows due to `smooth_n` and `diffs_n`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To be super clear lets look at the first few rows of training data for a specific metric before and after preprocessing. \n",
    "\n",
    "**Note**: Look at the last `time_idx` to see how the featurization works for a specific timestamp of data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "raw data\n"
     ]
    },
    {
     "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>system.cpu|user</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604474576</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474577</th>\n",
       "      <td>0.505050</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474578</th>\n",
       "      <td>1.005025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474579</th>\n",
       "      <td>0.502513</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474580</th>\n",
       "      <td>0.251889</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474581</th>\n",
       "      <td>0.251256</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474582</th>\n",
       "      <td>0.755668</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474583</th>\n",
       "      <td>1.005025</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474584</th>\n",
       "      <td>0.503778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474585</th>\n",
       "      <td>0.501253</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474586</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474587</th>\n",
       "      <td>0.753769</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|user\n",
       "time_idx                   \n",
       "1604474576         0.500000\n",
       "1604474577         0.505050\n",
       "1604474578         1.005025\n",
       "1604474579         0.502513\n",
       "1604474580         0.251889\n",
       "1604474581         0.251256\n",
       "1604474582         0.755668\n",
       "1604474583         1.005025\n",
       "1604474584         0.503778\n",
       "1604474585         0.501253\n",
       "1604474586         0.500000\n",
       "1604474587         0.753769"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "metric = 'system.cpu|user'\n",
    "print('raw data')\n",
    "display(df_train[df_train.columns[df_train.columns.str.startswith(metric)]].head(3 + lags_n + smooth_n + diffs_n))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "featurized data\n"
     ]
    },
    {
     "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>system.cpu|user_lag0</th>\n",
       "      <th>system.cpu|user_lag1</th>\n",
       "      <th>system.cpu|user_lag2</th>\n",
       "      <th>system.cpu|user_lag3</th>\n",
       "      <th>system.cpu|user_lag4</th>\n",
       "      <th>system.cpu|user_lag5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604474588</th>\n",
       "      <td>0.084805</td>\n",
       "      <td>0.08333</td>\n",
       "      <td>-0.168342</td>\n",
       "      <td>-0.084805</td>\n",
       "      <td>0.084174</td>\n",
       "      <td>0.251045</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|user_lag0  system.cpu|user_lag1  system.cpu|user_lag2  \\\n",
       "time_idx                                                                       \n",
       "1604474588              0.084805               0.08333             -0.168342   \n",
       "\n",
       "            system.cpu|user_lag3  system.cpu|user_lag4  system.cpu|user_lag5  \n",
       "time_idx                                                                      \n",
       "1604474588             -0.084805              0.084174              0.251045  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('featurized data')\n",
    "display(df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(metric)]].head(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "manualy calculated\n"
     ]
    },
    {
     "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>system.cpu|user</th>\n",
       "      <th>diff</th>\n",
       "      <th>smoothed</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604474583</th>\n",
       "      <td>1.005025</td>\n",
       "      <td>0.249358</td>\n",
       "      <td>0.251045</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474584</th>\n",
       "      <td>0.503778</td>\n",
       "      <td>-0.501247</td>\n",
       "      <td>0.084174</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474585</th>\n",
       "      <td>0.501253</td>\n",
       "      <td>-0.002525</td>\n",
       "      <td>-0.084805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474586</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>-0.001253</td>\n",
       "      <td>-0.168342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604474587</th>\n",
       "      <td>0.753769</td>\n",
       "      <td>0.253769</td>\n",
       "      <td>0.083330</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|user      diff  smoothed\n",
       "time_idx                                       \n",
       "1604474583         1.005025  0.249358  0.251045\n",
       "1604474584         0.503778 -0.501247  0.084174\n",
       "1604474585         0.501253 -0.002525 -0.084805\n",
       "1604474586         0.500000 -0.001253 -0.168342\n",
       "1604474587         0.753769  0.253769  0.083330"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print('manualy calculated')\n",
    "df_manual_example = df_train[df_train.columns[df_train.columns.str.startswith(metric)]].copy()\n",
    "# take diff\n",
    "df_manual_example['diff'] = df_manual_example[metric].diff(diffs_n)\n",
    "# apply smoothing\n",
    "df_manual_example['smoothed'] = df_manual_example['diff'].rolling(smooth_n).mean()\n",
    "display(df_manual_example.head(3 + lags_n + smooth_n + diffs_n).tail(1 + smooth_n + diffs_n))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Above you can see how one raw metric value is now being preprocessed to be a vector of `lags_n` differenced and smoothed values. It is this matrix of smoothed differences that the model will use for both training and when making predictions. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So, for example, if a chart has 2 dimensions and we have set `lags_n` to be 5 then our featurized 'matrix' of numbers will be a 2*(1+5) matrix. In reality this matrix is just flattened into a feature vector of 2 * (1+5) = 12 floating point values. The cell below shows this for the `system.net` chart as that is an example with 2 dimensions. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 12)\n",
      "[[ 37.26846759  40.19627126 -30.70439911 -40.85616748   6.4435552\n",
      "   32.54618009  32.69499079 -12.82253647 -24.87193934 -14.78879801\n",
      "   38.33229574   7.02717463]]\n"
     ]
    }
   ],
   "source": [
    "# lets look at our first feature vector for the 'system.net' model \n",
    "chart = 'system.net'\n",
    "print(df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(chart)]].head(1).shape)\n",
    "print(df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(chart)]].head(1).values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train models"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have our preprocessed training data we will train a model for each chart using our featurized data that represents each time step for each chart as a differenced, smoothed, and lagged matrix for each chart."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train model for system.cpu using X_train of (14387, 54)\n",
      "train model for system.load using X_train of (14387, 18)\n",
      "train model for system.net using X_train of (14387, 12)\n",
      "train model for system.io using X_train of (14387, 12)\n",
      "train model for system.ip using X_train of (14387, 12)\n",
      "train model for system.idlejitter using X_train of (14387, 18)\n"
     ]
    }
   ],
   "source": [
    "# loop over each chart in scope and train a model for each\n",
    "for chart in charts_in_scope:\n",
    "    # pull out the columns relating to the chart based on what thier name startswith and put it into a numpy array of values\n",
    "    X_train = df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(chart)]].values\n",
    "    print(f'train model for {chart} using X_train of {X_train.shape}')\n",
    "    # call the fit() method on each initialized model and pass it the full numpy array of our featurized training data\n",
    "    models[chart] = models[chart].fit(X_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So we have now trained our models, one for each chart based on our preprocessed training data. To be concrete we will look at some example obvervations our model has been trained on. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "timestamp=1604474588\n",
      "feature vector for 0th training observation for system.cpu model:\n",
      "[ 0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      " -0.0835422   0.0837521   0.          0.0835422   0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.08480479  0.16708229  0.16582914 -0.00084172 -0.08333017 -0.0004219\n",
      "  0.08480479  0.08333017 -0.16834172 -0.08480479  0.084174    0.25104532]\n"
     ]
    }
   ],
   "source": [
    "# lets look at the first matrix or \"feature vector\" for our first model\n",
    "obs_n = 0\n",
    "model_n = 0\n",
    "print(f'timestamp={df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(charts_in_scope[model_n])]].index[obs_n]}')\n",
    "print(f'feature vector for {obs_n}th training observation for {charts_in_scope[model_n]} model:')\n",
    "print(df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(charts_in_scope[model_n])]].values[obs_n]) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "timestamp=1604474589\n",
      "feature vector for 1th training observation for system.cpu model:\n",
      "[ 0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      "  0.         -0.0835422   0.0837521   0.          0.0835422   0.\n",
      "  0.          0.          0.          0.          0.          0.\n",
      " -0.08144416  0.08480479  0.16708229  0.16582914 -0.00084172 -0.08333017\n",
      " -0.08270361  0.08480479  0.08333017 -0.16834172 -0.08480479  0.084174  ]\n"
     ]
    }
   ],
   "source": [
    "# and the next one\n",
    "obs_n = 1\n",
    "model_n = 0\n",
    "print(f'timestamp={df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(charts_in_scope[model_n])]].index[obs_n]}')\n",
    "print(f'feature vector for {obs_n}th training observation for {charts_in_scope[model_n]} model:')\n",
    "print(df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(charts_in_scope[model_n])]].values[obs_n]) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If you look close enough at the above two cells you will see the same values be shifted for each lag."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each matrix of numbers above _is_ the representation we give to our model of each time step. This is how the model views each chart - a matrix (\"feature vector\" to sound fancy) of floating point numbers encoding some differenced and smoothed information about the last `lags_n` observations for each dimension in the specific chart we are modeling. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: Within the anomalies collector, at some regular interval, as defined by `train_every_n` in the `anomalies.conf` file, we will repeat the above training step to retrain all models on the most recent window of available training data. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Vistualize Training Data & Training Anomaly Scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have covered what our feature vectors look like and trained our model. Lets see if we can visualize things a little to help our intuition of whats going on. \n",
    "\n",
    "To do this we will use our trained model to get back an anomaly score for each of our training observations. \n",
    "\n",
    "Then we will visualize a random sample of training data alongside the most highly scored training data (i.e most 'anomalous') and we should expect to see stark differences. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(14387, 54)\n",
      "(14387,)\n",
      "(40, 54)\n",
      "(20, 54)\n",
      "(60, 54)\n"
     ]
    }
   ],
   "source": [
    "# lets pick a chart/model to look at\n",
    "chart = 'system.cpu'\n",
    "\n",
    "# pick a number of random samples to look at\n",
    "n_sample_random = 40\n",
    "# look also at the top n most anomalously scored training data\n",
    "n_high_score = 20\n",
    "\n",
    "# get all the training data for that model\n",
    "X_train = df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(chart)]].values\n",
    "X_train_timestamps = df_train_processed[df_train_processed.columns[df_train_processed.columns.str.startswith(chart)]].index.values\n",
    "n_features = X_train.shape[1]\n",
    "\n",
    "# score all our training data using the decision_function() method of the trained model - this will give use the raw internal anomaly score of the trained model\n",
    "X_train_scores = models[chart].decision_function(X_train)\n",
    "\n",
    "# determine some useful indices we can use to pull out the data we want\n",
    "idx_random = np.random.randint(n_sample_random, size=n_sample_random)\n",
    "idx_high = np.argpartition(X_train_scores, -n_high_score)[-n_high_score:]\n",
    "idx = idx_random.tolist() + idx_high.tolist()\n",
    "\n",
    "print(X_train.shape)\n",
    "print(X_train_scores.shape)\n",
    "\n",
    "X_train_random = X_train[idx_random,:]\n",
    "print(X_train_random.shape)\n",
    "\n",
    "X_train_high = X_train[idx_high,:]\n",
    "print(X_train_high.shape)\n",
    "\n",
    "X_train_examples = X_train[idx,:]\n",
    "print(X_train_examples.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we will just plot the random sample of our feature vectors as a line for each observation. \n",
    "\n",
    "**Note**: The x-axis here does not really mean much, it will just be the most recent and last lags_n values for each dimension on the chart. So this line plot if just a way to visualize the raw data for comparing observations. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_axis = np.array([n for n in range(0, n_features)])\n",
    "plt.figure(figsize=(15,10))\n",
    "p1 = plt.plot(x_axis, X_train_random.T)\n",
    "plt.xlim(0, n_features)\n",
    "plt.title(f'Random sample of {n_sample_random} training data feature vectors.')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now lets plot the same plot as above, _**but**_ also include the feature vectors of the training observations that had this highest anomaly scores. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,10))\n",
    "p1 = plt.plot(x_axis, X_train_examples.T)\n",
    "plt.xlim(0, n_features)\n",
    "plt.title(f'Random sample of {n_sample_random} training data feature vectors & top {n_high_score} highest scored feature vectors in training data')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now in the above plot we can see that the highest scored feature vectors or 'line' that we added in seem to stand out and look much 'different' to how the plot did before.\n",
    "\n",
    "**Note**: Don't forget to look at how the y-axis is also very different between the two plots."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another way to show this is via a heatmap of our random training data stacked alongside our highly anomalous training data. SO the data above the line in the plot below is our random sample of observations while that below is our highly scored training data. \n",
    "\n",
    "You should be able to see a clear difference, even if its not very interpretable to us what the differences in numbers actually represents visually in terms of how it might look in the normal Netdata dashboard - we left that world behind a couple of cells back and are now looking at things in terms of how the model see's our data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1152 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(10, 16))\n",
    "ax.imshow(X_train_examples, interpolation='nearest', cmap='Reds')\n",
    "plt.title(f'Feature heatmap of random sample of {n_sample_random} random training observations & {n_high_score} most anomalous training observations')\n",
    "plt.plot([n_sample_random for n in range(1,n_features+1)])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets pick the top few anomalies from the training data and highlight them in the Netdata dashboard to see how things look in the normal dashboard view."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(0) http://london.my-netdata.io/#menu_system_submenu_cpu;after=1604480183000;before=1604480503000;highlight_after=1604480473000;highlight_before=1604480493000\n",
      "(1) http://london.my-netdata.io/#menu_system_submenu_cpu;after=1604480181000;before=1604480501000;highlight_after=1604480471000;highlight_before=1604480491000\n",
      "(2) http://london.my-netdata.io/#menu_system_submenu_cpu;after=1604480180000;before=1604480500000;highlight_after=1604480470000;highlight_before=1604480490000\n",
      "(3) http://london.my-netdata.io/#menu_system_submenu_cpu;after=1604480182000;before=1604480502000;highlight_after=1604480472000;highlight_before=1604480492000\n",
      "(4) http://london.my-netdata.io/#menu_system_submenu_cpu;after=1604479154000;before=1604479474000;highlight_after=1604479444000;highlight_before=1604479464000\n"
     ]
    }
   ],
   "source": [
    "for n in range(5):\n",
    "    if chart == 'system.net':\n",
    "        submenu = '#menu_system_submenu_network'\n",
    "    elif chart == 'system.ip':\n",
    "        submenu = '#menu_system_submenu_network'\n",
    "    elif chart == 'system.cpu':\n",
    "        submenu = '#menu_system_submenu_cpu'\n",
    "    elif chart == 'system.load':\n",
    "        submenu = '#menu_system_submenu_load'\n",
    "    elif chart == 'system.io':\n",
    "        submenu = '#menu_system_submenu_disk'\n",
    "    elif chart == 'system.idlejitter':\n",
    "        submenu = '#menu_system_submenu_idlejitter'\n",
    "    after = (X_train_timestamps[idx_high][n]*1000)-(60000*5)\n",
    "    before = (X_train_timestamps[idx_high][n]*1000)+(10000*2)\n",
    "    highlight_after = (X_train_timestamps[idx_high][n]*1000)-10000\n",
    "    highlight_before = (X_train_timestamps[idx_high][n]*1000)+10000\n",
    "    print(f'({n}) http://{host}/{submenu};after={after};before={before};highlight_after={highlight_after};highlight_before={highlight_before}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get pediction data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have our trained models for each chart we can use them in looking at incoming observations and 'ask' the trained models how 'unusual' it thinks they are. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "we now have 10 recent preprocessed feature vectors to predict on.\n"
     ]
    }
   ],
   "source": [
    "# define an empty dataframe we can store enough recent data into to generate our feature vector's for recent data on\n",
    "df_recent = pd.DataFrame()\n",
    "\n",
    "# simulate n_prediction_steps of getting latest data, making feature vecotr and getting predicitons\n",
    "for prediction_step in range(n_prediction_steps):\n",
    "    time.sleep(1) # sleep for a second to ensure getting a new timestamp from the host\n",
    "    df_latest = get_allmetrics(host=host, charts=charts_in_scope, wide=True)[df_train.columns]\n",
    "    df_latest['time_idx'] = int(time.time())\n",
    "    df_latest = df_latest.set_index('time_idx')\n",
    "    # just keep enough recent data to generate each feature vector\n",
    "    df_recent = df_recent.append(df_latest).tail((lags_n + smooth_n + diffs_n) * 2)\n",
    "    \n",
    "    # now lets featurize our recent data to be able to get predictions from the model for each observation\n",
    "    df_predict_processed = make_features(df_recent, lags_n, diffs_n, smooth_n)\n",
    "\n",
    "print(f'we now have {df_predict_processed.shape[0]} recent preprocessed feature vectors to predict on.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10, 126)\n"
     ]
    },
    {
     "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>system.cpu|guest_lag0</th>\n",
       "      <th>system.cpu|guest_lag1</th>\n",
       "      <th>system.cpu|guest_lag2</th>\n",
       "      <th>system.cpu|guest_lag3</th>\n",
       "      <th>system.cpu|guest_lag4</th>\n",
       "      <th>system.cpu|guest_lag5</th>\n",
       "      <th>system.cpu|guest_nice_lag0</th>\n",
       "      <th>system.cpu|guest_nice_lag1</th>\n",
       "      <th>system.cpu|guest_nice_lag2</th>\n",
       "      <th>system.cpu|guest_nice_lag3</th>\n",
       "      <th>...</th>\n",
       "      <th>system.net|received_lag2</th>\n",
       "      <th>system.net|received_lag3</th>\n",
       "      <th>system.net|received_lag4</th>\n",
       "      <th>system.net|received_lag5</th>\n",
       "      <th>system.net|sent_lag0</th>\n",
       "      <th>system.net|sent_lag1</th>\n",
       "      <th>system.net|sent_lag2</th>\n",
       "      <th>system.net|sent_lag3</th>\n",
       "      <th>system.net|sent_lag4</th>\n",
       "      <th>system.net|sent_lag5</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>time_idx</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1604488989</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>20.950889</td>\n",
       "      <td>-48.192025</td>\n",
       "      <td>-15.095592</td>\n",
       "      <td>51.158781</td>\n",
       "      <td>33.586547</td>\n",
       "      <td>13.133732</td>\n",
       "      <td>42.412212</td>\n",
       "      <td>-15.099732</td>\n",
       "      <td>26.433434</td>\n",
       "      <td>-54.686425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604488991</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>30.414337</td>\n",
       "      <td>20.950889</td>\n",
       "      <td>-48.192025</td>\n",
       "      <td>-15.095592</td>\n",
       "      <td>2.353946</td>\n",
       "      <td>33.586547</td>\n",
       "      <td>13.133732</td>\n",
       "      <td>42.412212</td>\n",
       "      <td>-15.099732</td>\n",
       "      <td>26.433434</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604488992</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>1.330481</td>\n",
       "      <td>30.414337</td>\n",
       "      <td>20.950889</td>\n",
       "      <td>-48.192025</td>\n",
       "      <td>-93.153763</td>\n",
       "      <td>2.353946</td>\n",
       "      <td>33.586547</td>\n",
       "      <td>13.133732</td>\n",
       "      <td>42.412212</td>\n",
       "      <td>-15.099732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604488993</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-72.029869</td>\n",
       "      <td>1.330481</td>\n",
       "      <td>30.414337</td>\n",
       "      <td>20.950889</td>\n",
       "      <td>-1.209190</td>\n",
       "      <td>-93.153763</td>\n",
       "      <td>2.353946</td>\n",
       "      <td>33.586547</td>\n",
       "      <td>13.133732</td>\n",
       "      <td>42.412212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1604488994</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-12.498062</td>\n",
       "      <td>-72.029869</td>\n",
       "      <td>1.330481</td>\n",
       "      <td>30.414337</td>\n",
       "      <td>-23.862861</td>\n",
       "      <td>-1.209190</td>\n",
       "      <td>-93.153763</td>\n",
       "      <td>2.353946</td>\n",
       "      <td>33.586547</td>\n",
       "      <td>13.133732</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 126 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            system.cpu|guest_lag0  system.cpu|guest_lag1  \\\n",
       "time_idx                                                   \n",
       "1604488989                    0.0                    0.0   \n",
       "1604488991                    0.0                    0.0   \n",
       "1604488992                    0.0                    0.0   \n",
       "1604488993                    0.0                    0.0   \n",
       "1604488994                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_lag2  system.cpu|guest_lag3  \\\n",
       "time_idx                                                   \n",
       "1604488989                    0.0                    0.0   \n",
       "1604488991                    0.0                    0.0   \n",
       "1604488992                    0.0                    0.0   \n",
       "1604488993                    0.0                    0.0   \n",
       "1604488994                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_lag4  system.cpu|guest_lag5  \\\n",
       "time_idx                                                   \n",
       "1604488989                    0.0                    0.0   \n",
       "1604488991                    0.0                    0.0   \n",
       "1604488992                    0.0                    0.0   \n",
       "1604488993                    0.0                    0.0   \n",
       "1604488994                    0.0                    0.0   \n",
       "\n",
       "            system.cpu|guest_nice_lag0  system.cpu|guest_nice_lag1  \\\n",
       "time_idx                                                             \n",
       "1604488989                         0.0                         0.0   \n",
       "1604488991                         0.0                         0.0   \n",
       "1604488992                         0.0                         0.0   \n",
       "1604488993                         0.0                         0.0   \n",
       "1604488994                         0.0                         0.0   \n",
       "\n",
       "            system.cpu|guest_nice_lag2  system.cpu|guest_nice_lag3  ...  \\\n",
       "time_idx                                                            ...   \n",
       "1604488989                         0.0                         0.0  ...   \n",
       "1604488991                         0.0                         0.0  ...   \n",
       "1604488992                         0.0                         0.0  ...   \n",
       "1604488993                         0.0                         0.0  ...   \n",
       "1604488994                         0.0                         0.0  ...   \n",
       "\n",
       "            system.net|received_lag2  system.net|received_lag3  \\\n",
       "time_idx                                                         \n",
       "1604488989                 20.950889                -48.192025   \n",
       "1604488991                 30.414337                 20.950889   \n",
       "1604488992                  1.330481                 30.414337   \n",
       "1604488993                -72.029869                  1.330481   \n",
       "1604488994                -12.498062                -72.029869   \n",
       "\n",
       "            system.net|received_lag4  system.net|received_lag5  \\\n",
       "time_idx                                                         \n",
       "1604488989                -15.095592                 51.158781   \n",
       "1604488991                -48.192025                -15.095592   \n",
       "1604488992                 20.950889                -48.192025   \n",
       "1604488993                 30.414337                 20.950889   \n",
       "1604488994                  1.330481                 30.414337   \n",
       "\n",
       "            system.net|sent_lag0  system.net|sent_lag1  system.net|sent_lag2  \\\n",
       "time_idx                                                                       \n",
       "1604488989             33.586547             13.133732             42.412212   \n",
       "1604488991              2.353946             33.586547             13.133732   \n",
       "1604488992            -93.153763              2.353946             33.586547   \n",
       "1604488993             -1.209190            -93.153763              2.353946   \n",
       "1604488994            -23.862861             -1.209190            -93.153763   \n",
       "\n",
       "            system.net|sent_lag3  system.net|sent_lag4  system.net|sent_lag5  \n",
       "time_idx                                                                      \n",
       "1604488989            -15.099732             26.433434            -54.686425  \n",
       "1604488991             42.412212            -15.099732             26.433434  \n",
       "1604488992             13.133732             42.412212            -15.099732  \n",
       "1604488993             33.586547             13.133732             42.412212  \n",
       "1604488994              2.353946             33.586547             13.133732  \n",
       "\n",
       "[5 rows x 126 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(df_predict_processed.shape)\n",
    "df_predict_processed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above featurized prediction data should be identical in terms of structure and schema to the featurized training data we explored above. This is what is expected by the model.  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Get predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488989\n",
      "model=system.cpu, anomaly_probability=0.0291, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0822, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.007, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0102, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0527, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488991\n",
      "model=system.cpu, anomaly_probability=0.0355, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0658, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0082, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0077, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0563, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488992\n",
      "model=system.cpu, anomaly_probability=0.0416, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0438, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.013, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0143, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0489, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488993\n",
      "model=system.cpu, anomaly_probability=0.0386, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0068, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0153, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0138, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.034, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488994\n",
      "model=system.cpu, anomaly_probability=0.0329, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0078, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0152, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0124, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0307, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488995\n",
      "model=system.cpu, anomaly_probability=0.0347, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.006, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0119, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0062, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0275, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488996\n",
      "model=system.cpu, anomaly_probability=0.0356, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0036, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0051, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0051, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0304, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488998\n",
      "model=system.cpu, anomaly_probability=0.031, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0012, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0068, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0119, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0241, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604488999\n",
      "model=system.cpu, anomaly_probability=0.023, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0012, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0121, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0163, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0198, anomaly_flag=0\n",
      "----------------------------------------------------------------------------------------------------\n",
      "predictions for time 1604489000\n",
      "model=system.cpu, anomaly_probability=0.0226, anomaly_flag=0\n",
      "model=system.load, anomaly_probability=0.0018, anomaly_flag=0\n",
      "model=system.net, anomaly_probability=0.0171, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0014, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0193, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0184, anomaly_flag=0\n"
     ]
    }
   ],
   "source": [
    "# for each recent feature vector, get a prediction\n",
    "for time_idx, row in df_predict_processed.iterrows():\n",
    "    \n",
    "    print('-'*100)\n",
    "    print(f'predictions for time {time_idx}')\n",
    "    \n",
    "    # convert our row into the expected 'flattened' feature vector\n",
    "    df_tmp = row.to_frame().transpose()\n",
    "    \n",
    "    for model in models:\n",
    "        \n",
    "        # pull out relevant array of features for the model in question\n",
    "        X_predict = df_tmp[df_tmp.columns[df_tmp.columns.str.startswith(model)]].values\n",
    "        \n",
    "        # call the predict_proba() and predict() methods on the trained data in order to make a prediction\n",
    "        anomaly_probability = round(models[model].predict_proba(X_predict)[-1][1],4)\n",
    "        anomaly_flag = models[model].predict(X_predict)[-1]\n",
    "        \n",
    "        print(f'model={model}, anomaly_probability={anomaly_probability}, anomaly_flag={anomaly_flag}')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the above we should generally see low `anomaly_probability` values (assuming nothing has blown up on the host you used between the time you ran the training cells above and the predictions above). "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets just do one last little thing to try show what is going on here and why we put so much effort and focus into the featurization above.\n",
    "\n",
    "We will take some of the feature vectors we predicted on for each model, randomly shuffle the values around so as to make an unusual looking observation, and see what sort of an anomaly probability that gives us. (hint: it should be higher then those above :) )."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "------------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "model=system.cpu, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.load, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.net, anomaly_probability=0.0119, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0058, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0085, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0248, anomaly_flag=0\n",
      "------------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "model=system.cpu, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.load, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.net, anomaly_probability=0.0063, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0063, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0103, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.1029, anomaly_flag=0\n",
      "------------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "model=system.cpu, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.load, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.net, anomaly_probability=0.0093, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0028, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0099, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.1332, anomaly_flag=0\n",
      "------------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "model=system.cpu, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.load, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.net, anomaly_probability=0.0128, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0053, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0064, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.0155, anomaly_flag=0\n",
      "------------------------------------------------------------------------------------------------------------------------------------------------------\n",
      "model=system.cpu, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.load, anomaly_probability=1.0, anomaly_flag=1\n",
      "model=system.net, anomaly_probability=0.0074, anomaly_flag=0\n",
      "model=system.io, anomaly_probability=0.0138, anomaly_flag=0\n",
      "model=system.ip, anomaly_probability=0.0078, anomaly_flag=0\n",
      "model=system.idlejitter, anomaly_probability=0.2852, anomaly_flag=0\n"
     ]
    }
   ],
   "source": [
    "# sample some prediction data and shuffle values\n",
    "n_predict_sample = 5\n",
    "X_predict_sample = df_predict_processed.sample(n_predict_sample).values.T\n",
    "np.random.shuffle(X_predict_sample)\n",
    "X_predict_sample = X_predict_sample.T\n",
    "df_predict_shuffled = pd.DataFrame(X_predict_sample, columns=df_predict_processed.columns)\n",
    "for row in df_predict_shuffled.index:\n",
    "    print('-'*150)\n",
    "    for model in models:\n",
    "        X_predict = df_predict_shuffled.loc[row:row][df_predict_shuffled.columns[df_predict_shuffled.columns.str.startswith(model)]].values\n",
    "        anomaly_probability = round(models[model].predict_proba(X_predict)[-1][1],4)\n",
    "        anomaly_flag = models[model].predict(X_predict)[-1]\n",
    "        print(f'model={model}, anomaly_probability={anomaly_probability}, anomaly_flag={anomaly_flag}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We _should_ see some higher anomaly probabilities above than in the actual real predictions we had previously made. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## But what _is_ the model?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To try and lift the lid a little on what the model actually is and how it is calculating anomaly probabilities lets take a look at one trained model and what it actually is. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model for chart system.cpu:\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'contamination': 0.001,\n",
       " 'n_components': 2,\n",
       " 'n_selected_components': 2,\n",
       " 'copy': True,\n",
       " 'whiten': False,\n",
       " 'svd_solver': 'auto',\n",
       " 'tol': 0.0,\n",
       " 'iterated_power': 'auto',\n",
       " 'random_state': None,\n",
       " 'weighted': True,\n",
       " 'standardization': True,\n",
       " '_classes': 2,\n",
       " 'scaler_': StandardScaler(),\n",
       " 'detector_': PCA(n_components=2),\n",
       " 'n_components_': 2,\n",
       " 'components_': array([[ 2.80082352e-18,  3.03794228e-18, -2.70731444e-19,\n",
       "         -3.92363821e-19,  1.48369551e-19, -3.63283111e-19,\n",
       "         -4.39041383e-22, -8.31401705e-21,  1.39900394e-20,\n",
       "          1.30685531e-20, -9.11087543e-21,  0.00000000e+00,\n",
       "         -2.43885139e-01, -3.29203940e-01, -1.73697634e-01,\n",
       "          1.29799362e-01,  3.23636716e-01,  2.79958132e-01,\n",
       "          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
       "          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
       "         -1.51045154e-01, -8.76462672e-02,  6.10211833e-02,\n",
       "          1.98659891e-01,  2.27264690e-01,  1.49883841e-01,\n",
       "         -1.12372782e-01, -9.74261497e-02, -3.51807526e-02,\n",
       "          8.00901689e-02,  1.02115746e-01,  7.26665255e-02,\n",
       "         -1.24195520e-03, -1.22739202e-02, -1.72542528e-02,\n",
       "         -3.10527251e-03,  6.40218609e-03,  1.38531055e-02,\n",
       "         -2.85767294e-01, -2.95111985e-01, -1.27987592e-01,\n",
       "          1.47699610e-01,  3.13336131e-01,  2.93640283e-01,\n",
       "         -4.51371443e-02, -6.59597548e-02, -7.06319548e-02,\n",
       "         -1.24793697e-02,  2.99234105e-02,  6.52240921e-02],\n",
       "        [-1.60824947e-17, -1.14827710e-17,  9.61490236e-19,\n",
       "         -3.01121753e-19,  9.15825774e-20, -1.79272007e-19,\n",
       "         -3.09077873e-20, -2.81027414e-20, -4.26944492e-21,\n",
       "          1.35805909e-20, -1.02490202e-20, -0.00000000e+00,\n",
       "          9.19937111e-02, -1.01829472e-01, -2.77808201e-01,\n",
       "         -2.74783232e-01, -9.60873349e-02,  9.77127933e-02,\n",
       "         -0.00000000e+00, -0.00000000e+00, -0.00000000e+00,\n",
       "         -0.00000000e+00, -0.00000000e+00, -0.00000000e+00,\n",
       "         -1.93270784e-01, -3.33926708e-01, -3.73609183e-01,\n",
       "         -2.84145201e-01, -1.33619732e-01, -1.36922292e-02,\n",
       "         -1.04312593e-03, -7.56532493e-02, -1.20059231e-01,\n",
       "         -9.40272399e-02, -2.46077516e-02,  3.28380477e-02,\n",
       "          1.53039332e-02,  9.98811847e-03, -1.30719454e-03,\n",
       "         -1.23054923e-02, -1.04461356e-02, -2.39789225e-03,\n",
       "         -4.56703551e-02, -2.42918118e-01, -3.80897606e-01,\n",
       "         -3.54023992e-01, -1.80861638e-01,  1.67598654e-02,\n",
       "          1.10389117e-02, -1.99933307e-02, -6.19464953e-02,\n",
       "         -1.01962973e-01, -9.31777410e-02, -5.72142476e-02]]),\n",
       " 'n_selected_components_': 2,\n",
       " 'w_components_': array([0.09782853, 0.08702647]),\n",
       " 'selected_components_': array([[ 2.80082352e-18,  3.03794228e-18, -2.70731444e-19,\n",
       "         -3.92363821e-19,  1.48369551e-19, -3.63283111e-19,\n",
       "         -4.39041383e-22, -8.31401705e-21,  1.39900394e-20,\n",
       "          1.30685531e-20, -9.11087543e-21,  0.00000000e+00,\n",
       "         -2.43885139e-01, -3.29203940e-01, -1.73697634e-01,\n",
       "          1.29799362e-01,  3.23636716e-01,  2.79958132e-01,\n",
       "          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
       "          0.00000000e+00,  0.00000000e+00,  0.00000000e+00,\n",
       "         -1.51045154e-01, -8.76462672e-02,  6.10211833e-02,\n",
       "          1.98659891e-01,  2.27264690e-01,  1.49883841e-01,\n",
       "         -1.12372782e-01, -9.74261497e-02, -3.51807526e-02,\n",
       "          8.00901689e-02,  1.02115746e-01,  7.26665255e-02,\n",
       "         -1.24195520e-03, -1.22739202e-02, -1.72542528e-02,\n",
       "         -3.10527251e-03,  6.40218609e-03,  1.38531055e-02,\n",
       "         -2.85767294e-01, -2.95111985e-01, -1.27987592e-01,\n",
       "          1.47699610e-01,  3.13336131e-01,  2.93640283e-01,\n",
       "         -4.51371443e-02, -6.59597548e-02, -7.06319548e-02,\n",
       "         -1.24793697e-02,  2.99234105e-02,  6.52240921e-02],\n",
       "        [-1.60824947e-17, -1.14827710e-17,  9.61490236e-19,\n",
       "         -3.01121753e-19,  9.15825774e-20, -1.79272007e-19,\n",
       "         -3.09077873e-20, -2.81027414e-20, -4.26944492e-21,\n",
       "          1.35805909e-20, -1.02490202e-20, -0.00000000e+00,\n",
       "          9.19937111e-02, -1.01829472e-01, -2.77808201e-01,\n",
       "         -2.74783232e-01, -9.60873349e-02,  9.77127933e-02,\n",
       "         -0.00000000e+00, -0.00000000e+00, -0.00000000e+00,\n",
       "         -0.00000000e+00, -0.00000000e+00, -0.00000000e+00,\n",
       "         -1.93270784e-01, -3.33926708e-01, -3.73609183e-01,\n",
       "         -2.84145201e-01, -1.33619732e-01, -1.36922292e-02,\n",
       "         -1.04312593e-03, -7.56532493e-02, -1.20059231e-01,\n",
       "         -9.40272399e-02, -2.46077516e-02,  3.28380477e-02,\n",
       "          1.53039332e-02,  9.98811847e-03, -1.30719454e-03,\n",
       "         -1.23054923e-02, -1.04461356e-02, -2.39789225e-03,\n",
       "         -4.56703551e-02, -2.42918118e-01, -3.80897606e-01,\n",
       "         -3.54023992e-01, -1.80861638e-01,  1.67598654e-02,\n",
       "          1.10389117e-02, -1.99933307e-02, -6.19464953e-02,\n",
       "         -1.01962973e-01, -9.31777410e-02, -5.72142476e-02]]),\n",
       " 'selected_w_components_': array([0.09782853, 0.08702647]),\n",
       " 'decision_scores_': array([68.8878554 , 63.17257343, 67.21956501, ..., 75.31682611,\n",
       "        76.71194073, 74.30143433]),\n",
       " 'threshold_': 1311.4654267524904,\n",
       " 'labels_': array([0, 0, 0, ..., 0, 0, 0]),\n",
       " '_mu': 89.38196355699694,\n",
       " '_sigma': 97.22056202996157}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# lets pick our first trained model\n",
    "model_to_explore = charts_in_scope[0]\n",
    "\n",
    "print(f'model for chart {model_to_explore}:')\n",
    "models[model_to_explore].__dict__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So above we see the various objects, mostly array's of numbers, that make up the internal state of the trained model. \n",
    "\n",
    "**Note**: for different types of models you will see different things in here. The below cells all assume `model='pca'` to walk through the example of how the default PCA model calculates an anomaly probability. Also for simplicity when we initialized the PCA model we set `n_selected_components=2` to make the calculations below easier to follow by telling PyOD to just use the first 2 principle components when calculating the anomaly scores. In the actual anomalies collector we use all the principle components.\n",
    "\n",
    "For PCA the main things of relevance below will be: \n",
    "\n",
    "- `selected_components_`: The actual principle components we want to use when calculating the anomaly score (aka `decision_score`).\n",
    "- `selected_w_components_`: The weights applied to each selected component, the first few typically matter more as they capture most of the variance in the original training data. \n",
    "- `decision_scores_`: The raw anomaly scores on all of the training observations, used to convert the anomaly score into something that more looks like an anomaly probability. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Lets look at the 'training' source code"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets look at what the PCA model does when it trains a model. W can see from the below that most of what is going on is fitting a [PCA from Sklearn](https://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html) to the training data `X`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\u001b[1;31mSignature:\u001b[0m \u001b[0mPCA\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mNone\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mSource:\u001b[0m   \n",
       "    \u001b[1;32mdef\u001b[0m \u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mNone\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;34m\"\"\"Fit detector. y is ignored in unsupervised methods.\n",
       "\n",
       "        Parameters\n",
       "        ----------\n",
       "        X : numpy array of shape (n_samples, n_features)\n",
       "            The input samples.\n",
       "\n",
       "        y : Ignored\n",
       "            Not used, present for API consistency by convention.\n",
       "\n",
       "        Returns\n",
       "        -------\n",
       "        self : object\n",
       "            Fitted estimator.\n",
       "        \"\"\"\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# validate inputs X and y (optional)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mX\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_set_n_classes\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# PCA is recommended to use on the standardized data (zero mean and\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# unit variance).\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mstandardization\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mscaler_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mstandardizer\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mkeep_scalar\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdetector_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0msklearn_PCA\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mn_components\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0mcopy\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0mwhiten\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mwhiten\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0msvd_solver\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msvd_solver\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0mtol\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtol\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0miterated_power\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0miterated_power\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                     \u001b[0mrandom_state\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrandom_state\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdetector_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# copy the attributes from the sklearn PCA object\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdetector_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components_\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcomponents_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdetector_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcomponents_\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# validate the number of components to be used for outlier detection\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components_\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mcheck_parameter\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components_\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                        \u001b[0minclude_left\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minclude_right\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                        \u001b[0mparam_name\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'n_selected_components_'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# use eigenvalues as the weights of eigenvectors\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mw_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mones\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_components_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mweighted\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mw_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdetector_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexplained_variance_ratio_\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# outlier scores is the sum of the weighted distances between each\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# sample to the eigenvectors. The eigenvectors with smaller\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# eigenvalues have more influence\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# Not all eigenvectors are used, only n_selected_components_ smallest\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;31m# are used since they better reflect the variance change\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcomponents_\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                    \u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m \u001b[1;33m*\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components_\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_w_components_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mw_components_\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                                      \u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m \u001b[1;33m*\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_selected_components_\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecision_scores_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msum\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mcdist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_components_\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m/\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_w_components_\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0maxis\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mravel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_process_decision_scores\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mFile:\u001b[0m      c:\\users\\andre\\documents\\repos\\netdata-community\\netdata-agent-api\\netdata-pandas\\venv\\lib\\site-packages\\pyod\\models\\pca.py\n",
       "\u001b[1;31mType:\u001b[0m      function\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "PCA.fit??"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Lets look at the 'prediction' source code"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets look at the source code for generating the anomaly probabilities."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\u001b[1;31mSignature:\u001b[0m \u001b[0mPCA\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpredict_proba\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'linear'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mSource:\u001b[0m   \n",
       "    \u001b[1;32mdef\u001b[0m \u001b[0mpredict_proba\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmethod\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'linear'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;34m\"\"\"Predict the probability of a sample being outlier. Two approaches\n",
       "        are possible:\n",
       "\n",
       "        1. simply use Min-max conversion to linearly transform the outlier\n",
       "           scores into the range of [0,1]. The model must be\n",
       "           fitted first.\n",
       "        2. use unifying scores, see :cite:`kriegel2011interpreting`.\n",
       "\n",
       "        Parameters\n",
       "        ----------\n",
       "        X : numpy array of shape (n_samples, n_features)\n",
       "            The input samples.\n",
       "\n",
       "        method : str, optional (default='linear')\n",
       "            probability conversion method. It must be one of\n",
       "            'linear' or 'unify'.\n",
       "\n",
       "        Returns\n",
       "        -------\n",
       "        outlier_probability : numpy array of shape (n_samples,)\n",
       "            For each observation, tells whether or not\n",
       "            it should be considered as an outlier according to the\n",
       "            fitted model. Return the outlier probability, ranging\n",
       "            in [0,1].\n",
       "        \"\"\"\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mcheck_is_fitted\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;34m'decision_scores_'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'threshold_'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'labels_'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mtrain_scores\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecision_scores_\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mtest_scores\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecision_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mprobs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mzeros\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_classes\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mif\u001b[0m \u001b[0mmethod\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;34m'linear'\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mscaler\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mMinMaxScaler\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain_scores\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mscaler\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                \u001b[0mtest_scores\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mravel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mclip\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m1\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[1;32mreturn\u001b[0m \u001b[0mprobs\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32melif\u001b[0m \u001b[0mmethod\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;34m'unify'\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[1;31m# turn output into probability\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mpre_erf_score\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mtest_scores\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_mu\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m/\u001b[0m \u001b[1;33m(\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                    \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_sigma\u001b[0m \u001b[1;33m*\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msqrt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0merf_score\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0merf\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mpre_erf_score\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0merf_score\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mclip\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mravel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m1\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mprobs\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[1;32mreturn\u001b[0m \u001b[0mprobs\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[1;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmethod\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m                             \u001b[1;34m'is not a valid probability conversion method'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mFile:\u001b[0m      c:\\users\\andre\\documents\\repos\\netdata-community\\netdata-agent-api\\netdata-pandas\\venv\\lib\\site-packages\\pyod\\models\\base.py\n",
       "\u001b[1;31mType:\u001b[0m      function\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "PCA.predict_proba??"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see its getting scores from some `decision_function()` method, so lets look at that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\u001b[1;31mSignature:\u001b[0m \u001b[0mPCA\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecision_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mSource:\u001b[0m   \n",
       "    \u001b[1;32mdef\u001b[0m \u001b[0mdecision_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;34m\"\"\"Predict raw anomaly score of X using the fitted detector.\n",
       "\n",
       "        The anomaly score of an input sample is computed based on different\n",
       "        detector algorithms. For consistency, outliers are assigned with\n",
       "        larger anomaly scores.\n",
       "\n",
       "        Parameters\n",
       "        ----------\n",
       "        X : numpy array of shape (n_samples, n_features)\n",
       "            The training input samples. Sparse matrices are accepted only\n",
       "            if they are supported by the base estimator.\n",
       "\n",
       "        Returns\n",
       "        -------\n",
       "        anomaly_scores : numpy array of shape (n_samples,)\n",
       "            The anomaly score of the input samples.\n",
       "        \"\"\"\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mcheck_is_fitted\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;34m'components_'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'w_components_'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[0mX\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcheck_array\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mstandardization\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mX\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mscaler_\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\n",
       "\u001b[0m\u001b[1;33m\n",
       "\u001b[0m        \u001b[1;32mreturn\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msum\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0mcdist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_components_\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m/\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mselected_w_components_\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\n",
       "\u001b[0m            \u001b[0maxis\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mravel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
       "\u001b[1;31mFile:\u001b[0m      c:\\users\\andre\\documents\\repos\\netdata-community\\netdata-agent-api\\netdata-pandas\\venv\\lib\\site-packages\\pyod\\models\\pca.py\n",
       "\u001b[1;31mType:\u001b[0m      function\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "PCA.decision_function??"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see here what actually look like relativley straightforward calculations, so lets try step through them below. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Ok so lets step through that!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So the cells above show the PyOD code under the hood - lets step through and recreate the a predicted score, step by step."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets begin by getting our feature vector that we would like an anomaly probability for."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "feature vector\n",
      "(1, 54)\n",
      "[[ 0.          0.          0.          0.          0.          0.\n",
      "   0.          0.          0.          0.          0.          0.\n",
      "   0.          0.          0.          0.          0.          0.\n",
      "   0.          0.          0.          0.          0.          0.\n",
      "   0.          0.          0.          0.          0.          0.\n",
      "  -0.1675042   0.          0.          0.1675042  -0.08291873  0.\n",
      "   0.          0.         -0.08333333  0.          0.          0.08333333\n",
      "  -0.167924   -0.08270047 -0.0824958   0.         -0.16396713 -0.00252527\n",
      "  -0.00104953  0.00168347  0.08500837  0.0837521  -0.32918117 -0.00252527]]\n"
     ]
    }
   ],
   "source": [
    "from scipy.spatial.distance import cdist\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "# get our feature vector for a random observation from our prediction data\n",
    "X = df_predict_processed[df_predict_processed.columns[df_predict_processed.columns.str.startswith(model_to_explore)]].sample(1).values\n",
    "\n",
    "print('feature vector')\n",
    "print(X.shape)\n",
    "print(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets look at the principle components the model will use to calculate our anomaly score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "selected components learned by model\n",
      "(2, 54)\n",
      "[[ 2.80082352e-18  3.03794228e-18 -2.70731444e-19 -3.92363821e-19\n",
      "   1.48369551e-19 -3.63283111e-19 -4.39041383e-22 -8.31401705e-21\n",
      "   1.39900394e-20  1.30685531e-20 -9.11087543e-21  0.00000000e+00\n",
      "  -2.43885139e-01 -3.29203940e-01 -1.73697634e-01  1.29799362e-01\n",
      "   3.23636716e-01  2.79958132e-01  0.00000000e+00  0.00000000e+00\n",
      "   0.00000000e+00  0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      "  -1.51045154e-01 -8.76462672e-02  6.10211833e-02  1.98659891e-01\n",
      "   2.27264690e-01  1.49883841e-01 -1.12372782e-01 -9.74261497e-02\n",
      "  -3.51807526e-02  8.00901689e-02  1.02115746e-01  7.26665255e-02\n",
      "  -1.24195520e-03 -1.22739202e-02 -1.72542528e-02 -3.10527251e-03\n",
      "   6.40218609e-03  1.38531055e-02 -2.85767294e-01 -2.95111985e-01\n",
      "  -1.27987592e-01  1.47699610e-01  3.13336131e-01  2.93640283e-01\n",
      "  -4.51371443e-02 -6.59597548e-02 -7.06319548e-02 -1.24793697e-02\n",
      "   2.99234105e-02  6.52240921e-02]\n",
      " [-1.60824947e-17 -1.14827710e-17  9.61490236e-19 -3.01121753e-19\n",
      "   9.15825774e-20 -1.79272007e-19 -3.09077873e-20 -2.81027414e-20\n",
      "  -4.26944492e-21  1.35805909e-20 -1.02490202e-20 -0.00000000e+00\n",
      "   9.19937111e-02 -1.01829472e-01 -2.77808201e-01 -2.74783232e-01\n",
      "  -9.60873349e-02  9.77127933e-02 -0.00000000e+00 -0.00000000e+00\n",
      "  -0.00000000e+00 -0.00000000e+00 -0.00000000e+00 -0.00000000e+00\n",
      "  -1.93270784e-01 -3.33926708e-01 -3.73609183e-01 -2.84145201e-01\n",
      "  -1.33619732e-01 -1.36922292e-02 -1.04312593e-03 -7.56532493e-02\n",
      "  -1.20059231e-01 -9.40272399e-02 -2.46077516e-02  3.28380477e-02\n",
      "   1.53039332e-02  9.98811847e-03 -1.30719454e-03 -1.23054923e-02\n",
      "  -1.04461356e-02 -2.39789225e-03 -4.56703551e-02 -2.42918118e-01\n",
      "  -3.80897606e-01 -3.54023992e-01 -1.80861638e-01  1.67598654e-02\n",
      "   1.10389117e-02 -1.99933307e-02 -6.19464953e-02 -1.01962973e-01\n",
      "  -9.31777410e-02 -5.72142476e-02]]\n"
     ]
    }
   ],
   "source": [
    "print('selected components learned by model')\n",
    "selected_components = models[model_to_explore].selected_components_\n",
    "print(selected_components.shape)\n",
    "print(selected_components)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets look at the weighting that the model will apply to each component when working out a weighted distance that will form the score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "selected components weights set by model\n",
      "(2,)\n",
      "[0.09782853 0.08702647]\n"
     ]
    }
   ],
   "source": [
    "print('selected components weights set by model')\n",
    "selected_w_components = models[model_to_explore].selected_w_components_\n",
    "print(selected_w_components.shape)\n",
    "print(selected_w_components)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets just call `predict_proba(X)` to get the score we will try to recreate."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anomaly probability [p(anomaly), p(not anomaly)]\n",
      "(1, 2)\n",
      "[[0.96450375 0.03549625]]\n"
     ]
    }
   ],
   "source": [
    "print('anomaly probability [p(anomaly), p(not anomaly)]')\n",
    "anomaly_probability = models[model_to_explore].predict_proba(X)\n",
    "print(anomaly_probability.shape)\n",
    "print(anomaly_probability)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Step 1: Standardize the feature vector"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The first step is to standardize the data as this is a default common practice when fitting a PCA to data. \n",
    "\n",
    "This is the default here which is good as it means that if you define any [custom models](https://github.com/netdata/netdata/blob/master/collectors/python.d.plugin/anomalies#custom-models) in your `anomalies.conf` file, you dont need to worry about them not being on the same or similar scales (e.g. cpu % vs disk usage etc) as that is all taken care of internally by the PyOD model. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.00000000e+00  0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      "   0.00000000e+00  0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      "   0.00000000e+00  0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      "  -2.70067980e-10 -2.70067979e-10 -2.70067980e-10 -2.70067979e-10\n",
      "  -2.70067980e-10 -2.70067979e-10  0.00000000e+00  0.00000000e+00\n",
      "   0.00000000e+00  0.00000000e+00  0.00000000e+00  0.00000000e+00\n",
      "   2.40674662e-10  2.40674661e-10  2.40674662e-10  2.40674661e-10\n",
      "   2.40674662e-10  2.40674661e-10 -2.66664208e+00 -9.29120645e-05\n",
      "  -4.65713239e-07  2.66664268e+00 -1.32018223e+00 -9.29236062e-05\n",
      "  -6.02436866e-12 -6.02436881e-12 -1.55106650e+00 -6.02436888e-12\n",
      "  -6.02436872e-12  1.55106650e+00 -3.86095274e-01 -1.90187398e-01\n",
      "  -1.89755931e-01 -5.35431133e-05 -3.77037179e-01 -5.84708821e-03\n",
      "  -5.22050433e-03  8.09092270e-03  4.15752327e-01  4.09701175e-01\n",
      "  -1.61015895e+00 -1.24094202e-02]]\n"
     ]
    }
   ],
   "source": [
    "X_scaled = models[model_to_explore].scaler_.transform(X)\n",
    "print(X_scaled)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Step 2: Calculate distance from each selected component "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we want to work out the distance of our feature vector to each of the selected components. \n",
    "\n",
    "This is the core part of what is going on when calculating an anomaly score using this model. \n",
    "\n",
    "If we have a strange feature vector then its going to be something we have not really seen before and so will not fit well into the lower dimensional representation learned by the PCA model. Hence it will have a somewhat larger distance from the selected principle components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 2)\n",
      "[[4.92401167 4.99332981]]\n"
     ]
    }
   ],
   "source": [
    "distance = cdist(X_scaled, selected_components)\n",
    "print(distance.shape)\n",
    "print(distance)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Note**: These distance numbers are just numbers, where bigger means more distant, but on their own are hard to interpret it terms of what they mean. So the next step will be to try and go from the weighted distance score to an anomaly probability by comparing the anomay score with all those we saw within the training data. It's the training data that will be our yardstick for trying to say just how anomalous a new observation is. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Step 3: Use a weighted average distance as the anomaly score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will take a weighted average of the distances, where the weights for each component will, by default, be related to the amount of variance in the original training data that each component 'explained' or represented (we can see this in the PyOD code [here](https://github.com/yzhao062/pyod/blob/master/pyod/models/pca.py#L252))."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "anomaly score\n",
      "[107.7102246]\n"
     ]
    }
   ],
   "source": [
    "print('anomaly score')\n",
    "anomaly_score = np.sum(distance / selected_w_components, axis=1).ravel()\n",
    "print(anomaly_score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Step 4: Convert anomaly score into anomaly probabilities based on the anomaly scores of the training data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we will use the decision scores we calculated from basically running the prediction process back over the training data when we originally fit our model. \n",
    "\n",
    "We will use the distribution of scores in the training data to try and re-scale our raw anomaly score to look more like something that can pass as a probability.\n",
    "\n",
    "**Note**: Strictly speaking this \"anomaly probability\" is not really a \"real\" probability in the sense of being some sort of more formal or theoretical output from some probabilistic process we have statistically modeled directly. Rather, its just a sensible re-scaling of our raw score based on what we saw in the training data. So a high 'probability' here really just means an unusual value in reference to the training data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.96450375 0.03549625]]\n"
     ]
    }
   ],
   "source": [
    "# get the raw anomaly scores from the training data\n",
    "train_anomaly_scores = models[model_to_explore].decision_scores_\n",
    "\n",
    "# create empty array for probabilities to go into\n",
    "anomaly_probability_manual_calc = np.zeros([X.shape[0], int(models[model_to_explore]._classes)])\n",
    "\n",
    "# create a scaler to rescale raw anomaly score to look more like a probability and be on the 0, 1 range\n",
    "scaler = MinMaxScaler().fit(train_anomaly_scores.reshape(-1, 1))\n",
    "\n",
    "# transform anomaly score into a probability by rescaling it based on the training data and clipping at 1\n",
    "anomaly_probability_manual_calc[:, 1] = scaler.transform(anomaly_score.reshape(-1, 1)).ravel().clip(0, 1)\n",
    "\n",
    "# use 1 - p(anomaly) as p(not anomaly)\n",
    "anomaly_probability_manual_calc[:, 0] = 1 - anomaly_probability_manual_calc[:, 1]\n",
    "\n",
    "print(anomaly_probability_manual_calc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Do they match?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.96450375 0.03549625]]\n",
      "[[0.96450375 0.03549625]]\n",
      "[[ True  True]]\n"
     ]
    }
   ],
   "source": [
    "# do they match?\n",
    "print(anomaly_probability)\n",
    "print(anomaly_probability_manual_calc)\n",
    "print(anomaly_probability == anomaly_probability_manual_calc)\n",
    "assert np.sum(anomaly_probability == anomaly_probability_manual_calc) == 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "....phew, thats it! Go get yourself a coffee :)"
   ]
  },
  {
   "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": "3.8.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}