utils.py

from random import randint
import pandas as pd
import requests, zipfile
try:
    from StringIO import StringIO
except ImportError:
    from io import StringIO
import os

def download_data():
    if 'train_FD004.txt' not in os.listdir('data'):
        print('Downloading Data...')
        # Download the data
        r = requests.get("https://ti.arc.nasa.gov/c/6/", stream=True)
        z = zipfile.ZipFile(StringIO.StringIO(r.content))
        z.extractall('data')
    else:
        print('Using previously downloaded data')
    
def load_data(data_path):   
    operational_settings = ['operational_setting_{}'.format(i + 1) for i in range (3)]
    sensor_columns = ['sensor_measurement_{}'.format(i + 1) for i in range(26)]
    cols = ['engine_no', 'time_in_cycles'] + operational_settings + sensor_columns
    data = pd.read_csv(data_path, sep=' ', header=-1, names=cols)
    data = data.drop(cols[-5:], axis=1)
    data['index'] = data.index
    data.index = data['index']
    data['time'] = pd.date_range('1/1/2000', periods=data.shape[0], freq='600s')
    print('Loaded data with:\n{} Recordings\n{} Engines'.format(
        data.shape[0], len(data['engine_no'].unique())))
    print('21 Sensor Measurements\n3 Operational Settings')
    return data

def new_labels(data, labels):
    ct_ids = []
    ct_times = []
    ct_labels = []
    data = data.copy()
    data['RUL'] = labels
    gb = data.groupby(['engine_no'])
    for engine_no_df in gb:
        instances = engine_no_df[1].shape[0]
        r = randint(5, instances - 1)
        ct_ids.append(engine_no_df[1].iloc[r,:]['engine_no'])
        ct_times.append(engine_no_df[1].iloc[r,:]['time'])
        ct_labels.append(engine_no_df[1].iloc[r,:]['RUL'])
    ct = pd.DataFrame({'engine_no': ct_ids,
                       'cutoff_time': ct_times,
                       'RUL': ct_labels})
    ct = ct[['engine_no', 'cutoff_time', 'RUL']]
    ct.index = ct['engine_no']
    ct.index = ct.index.rename('index')
    return ct

def make_cutoff_times(data):
    gb = data.groupby(['engine_no'])
    labels = []


    for engine_no_df in gb:
        instances = engine_no_df[1].shape[0]
        label = [instances - i - 1 for i in range(instances)]
        labels += label
    
    return new_labels(data, labels)

def feature_importances(X, reg, feats=5):
    feature_imps = [(imp, X.columns[i]) 
                    for i, imp in enumerate(reg.feature_importances_)]
    feature_imps.sort()
    feature_imps.reverse()
    for i, f in enumerate(feature_imps[0:feats]):
        print('{}: {} [{:.3f}]'.format(i + 1, f[1], f[0]))
    print('-----\n')
    return [f[1] for f in feature_imps[:feats]]