{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Assessing building fire risk with Random Forest\n", "** *Combining different governmental datasets to prioritize fire department resources* **\n", "\n", "## Background\n", "\n", "Fire departments across the United States spend considerable resources on educational and prevention campaigns to stop building fires before they happen. Moving from a shotgun approach to a targeted one that considers the likelihood of fire incidents could significantly improve the efficacy of these interventions. To do this, machine learning algorithms built on data from a wide variety of sources could vastly improve how we perform against randomness. \n", "\n", "Using data provided by the city of Sioux Falls, this notebook walks through the steps of predicting fire risk for every address listed in the city with a random forest classifier algorithm written in Python, and trained on data coming from several different county and municipal departments.\n" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Import libraries" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "# Data analysis and visualization\n", "import numpy as np\n", "import pandas as pd\n", "import seaborn as sns\n", "import matplotlib.pyplot as plt\n", "\n", "# Interactive maps \n", "import folium\n", "from folium.plugins import HeatMap\n", "\n", "# Machine learning\n", "from sklearn.ensemble import RandomForestClassifier\n", "from sklearn.metrics import roc_curve, roc_auc_score, auc\n", "from sklearn.model_selection import cross_val_score, GridSearchCV\n", "from sklearn.feature_selection import SelectKBest, f_classif\n", "from sklearn.pipeline import Pipeline" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Load and describe the data" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\pandas\\io\\excel.py:520: UserWarning: The 'parse_dates=True' keyword of read_excel was provided without an 'index_col' keyword value.\n", " warn(\"The 'parse_dates=True' keyword of read_excel was provided\"\n" ] } ], "source": [ "# Load the data into Python\n", "fires_df = pd.read_csv('data/fires2.csv', parse_dates=[6])\n", "parcels_df = pd.read_excel('data/parcels2.xlsx', parse_dates=True)\n", "foreclosures_df = pd.read_excel('data/foreclosures.xlsx', parse_dates=True)\n", "rent_reg_df = pd.read_excel('data/rental_registration_permits.xlsx', parse_dates=True)\n", "utility_disconnects_df = pd.read_excel('data/utility_disconnects.xlsx', parse_dates=True)\n", "code_cases_df = pd.read_excel('data/code_cases.xlsx', parse_dates = True)\n", "minnehaha_tax_df = pd.read_excel('data/minnehaha_tax.xlsx', parse_dates = True)\n", "lincoln_tax_df = pd.read_excel('data/lincoln_tax.xlsx', parse_dates = True)\n", "crime_df = pd.read_excel('data/crime.xlsx', parse_dates = True)" ] }, { "cell_type": "code", "execution_count": 3, "metadata": { "scrolled": true }, "outputs": [ { "data": { "text/plain": [ "Building fire 1258\n", "Passenger vehicle fire 639\n", "Cooking fire, confined to container 244\n", "Grass fire 198\n", "Dumpster or other outside trash receptacle fire 183\n", "Outside rubbish, trash or waste fire 124\n", "Brush, or brush and grass mixture fire 94\n", "Special outside fire, other 91\n", "Outside equipment fire 80\n", "Fire, other 71\n", "Mobile property (vehicle) fire, other 71\n", "Natural vegetation fire, other 70\n", "Trash or rubbish fire, contained 64\n", "Outside rubbish fire, other 57\n", "Fire in mobile home used as fixed residence 56\n", "Road freight or transport vehicle fire 35\n", "Forest, woods or wildland fire 19\n", "Off-road vehicle or heavy equipment fire 16\n", "Fires in structures other than in a building 15\n", "Outside storage fire 10\n", "Cultivated grain or crop fire 7\n", "Chimney or flue fire, confined to chimney or flue 7\n", "Fire in portable building, fixed location 7\n", "Camper or recreational vehicle (RV) fire 6\n", "Outside gas or vapor combustion explosion 5\n", "Fuel burner/boiler malfunction, fire confined 5\n", "Commercial Compactor fire, confined to rubbish 3\n", "Self-propelled motor home or recreational vehicle 3\n", "Incinerator overload or malfunction, fire confined 3\n", "Cultivated vegetation, crop fire, other 3\n", "Fire in mobile property used as a fixed structure, other 3\n", "Fire in motor home, camper, recreational vehicle 2\n", "Outside mailbox fire 2\n", "Cultivated trees or nursery stock fire 2\n", "Water vehicle fire 2\n", "Aircraft fire 2\n", "Outside stationary compactor/compacted trash fire 1\n", "Garbage dump or sanitary landfill fire 1\n", "Construction or demolition landfill fire 1\n", "Name: type, dtype: int64" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "fires_df.type.value_counts()" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "# Create copys of DataFrames so if I mess something up I don't have to load up everything again\n", "building_fires = fires_df[fires_df['id2']== 111].copy()\n", "parcels = parcels_df.copy()\n", "foreclosures = foreclosures_df.copy()\n", "rent_reg = rent_reg_df.copy()\n", "utility_disconnects = utility_disconnects_df.copy()\n", "code_cases = code_cases_df.copy()\n", "minnehaha_tax = minnehaha_tax_df.copy()\n", "lincoln_tax = lincoln_tax_df.copy()\n", "crime = crime_df.copy()" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Index(['TAG', 'COUNTYID', 'ADDRESS', 'OWNNAME1', 'OWNNAME2', 'OWNADDRESS',\n", " 'OWNCITY', 'OWNSTATE', 'OWNZIP', 'OWNZIP2', 'SQFT', 'ACREAGE',\n", " 'FRONTFOOT', 'LEGAL', 'ADDITION', 'ADDITIONNU', 'PARHOUSE', 'PARHALF',\n", " 'PARPR', 'PARSTREET', 'PARTYPE', 'PARPD', 'UNITNUM', 'ACTIVITY',\n", " 'LANDUSE', 'NUMUNITS', 'COUNTY', 'LegalStart', 'GlobalID', 'created_us',\n", " 'created_da', 'last_edite', 'last_edi_1', 'ADDITIONPR', 'PARCEL_LOT',\n", " 'PARCEL_TRA', 'BlockDesig', 'FORM_PRIMA', 'FORM_ACCES', 'FORM_SIGNE',\n", " 'FORM_DATE', 'FORM_COMME', 'DEPARTMENT', 'PARCELTYPE', 'ZIPCODE',\n", " 'Shape_Leng', 'Shape_Area'],\n", " dtype='object')" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" } ], "source": [ "parcels.columns" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
ASSESSEDVALUEYEAR
count1.974000e+031982.000000
mean1.281877e+052011.934914
std8.889734e+042.875913
min0.000000e+002008.000000
25%8.446325e+042010.000000
50%1.136345e+052012.000000
75%1.453120e+052014.000000
max1.562129e+062019.000000
\n", "
" ], "text/plain": [ " ASSESSEDVALUE YEAR\n", "count 1.974000e+03 1982.000000\n", "mean 1.281877e+05 2011.934914\n", "std 8.889734e+04 2.875913\n", "min 0.000000e+00 2008.000000\n", "25% 8.446325e+04 2010.000000\n", "50% 1.136345e+05 2012.000000\n", "75% 1.453120e+05 2014.000000\n", "max 1.562129e+06 2019.000000" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "foreclosures.describe()" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "Int64Index: 1258 entries, 652 to 2318\n", "Data columns (total 10 columns):\n", "FID 1258 non-null int64\n", "Join_Count 1258 non-null int64\n", "TARGET_FID 1258 non-null int64\n", "id 1258 non-null int64\n", "id2 1258 non-null int64\n", "type 1258 non-null object\n", "date 1258 non-null datetime64[ns]\n", "lat 1258 non-null float64\n", "lon 1258 non-null float64\n", "ADDRESS 1258 non-null object\n", "dtypes: datetime64[ns](1), float64(2), int64(5), object(2)\n", "memory usage: 108.1+ KB\n" ] } ], "source": [ "building_fires.info()" ] }, { "cell_type": "code", "execution_count": 8, "metadata": { "scrolled": true }, "outputs": [ { "data": { "text/html": [ "
" ], "text/plain": [ "" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Visualizing the fire incidents in a map\n", "fire_map = folium.Map(location=[43.54, -96.72], zoom_start=14, tiles='Stamen Terrain')\n", "heat_df = building_fires[['lat', 'lon']]\n", "heat_df = heat_df.dropna(axis=0, subset=['lat','lon'])\n", "heat_data = [[row['lat'],row['lon']] for index, row in heat_df.iterrows()]\n", "HeatMap(heat_data).add_to(fire_map)\n", "fire_map" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Data wrangling" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [], "source": [ "def cleanActivity(x):\n", " if x == 11 or x == 12 or x == 13:\n", " x = 'SINGLE OR TWO RESIDENTIAL'\n", " elif x == 21 or x == 22 or x == 23 or x == 24 or x == 25:\n", " x = 'MULTIFAMILY'\n", " elif x in list(range(31,39)):\n", " x = 'OFFICE AND PUBLIC SERVICE'\n", " elif x in list(range(40,50)):\n", " x = 'INSTITUTIONAL'\n", " elif x in list(range(51, 54)):\n", " x = 'COMMERCIAL'\n", " elif x == 61 or x in list(range(63,70)):\n", " x = 'INDUSTRIAL'\n", " elif x == 62:\n", " x = 'AIRPORT'\n", " elif x in range(70,98) or x in range(0, 9):\n", " x = 'OPEN SPACES/NA'\n", " return x\n", "\n", "\n", "parcels['ACTIVITY'] = parcels.ACTIVITY.apply(cleanActivity)\n", "\n", "# Parcels dataset\n", "df = pd.concat([parcels, pd.get_dummies(parcels.ACTIVITY, prefix='ACT_'), pd.get_dummies(parcels.PARPR), pd.get_dummies(parcels.PARCELTYPE), pd.get_dummies(parcels.PARTYPE), pd.get_dummies(parcels.COUNTY)], axis = 1)\n", "df.PARHOUSE.fillna(method='bfill', inplace = True)\n", "df = df[df.ACTIVITY != 'AIRPORT']\n", "df = df[df.ACTIVITY != 'OPEN SPACES/NA']\n", "parcels.drop(columns = ['BlockDesig', 'FORM_PRIMA', 'PARCELTYPE', 'FORM_COMME', 'PARCEL_TRA', 'PARCEL_LOT', 'created_us', 'TAG', 'ADDITION', 'ADDITIONPR', 'COUNTYID', 'OWNNAME1', 'OWNZIP', 'OWNZIP2', 'OWNNAME2', 'OWNCITY', 'OWNSTATE', 'OWNADDRESS', 'LEGAL', 'ADDITIONNU', 'PARHALF', 'PARPD', 'UNITNUM', 'GlobalID', 'last_edite', 'Shape_Leng', 'Shape_Area', 'DEPARTMENT', 'FORM_ACCES', 'FORM_SIGNE', 'FORM_DATE', 'created_da', 'last_edi_1', 'PARPR', 'PARTYPE', 'COUNTY', 'PARSTREET', 'LegalStart', 'ACTIVITY'], inplace = True)\n", "df.drop(columns = ['BlockDesig', 'ACREAGE', 'FORM_PRIMA', 'PARCELTYPE', 'FORM_COMME', 'PARCEL_TRA', 'PARCEL_LOT', 'created_us', 'TAG', 'ADDITION', 'ADDITIONPR', 'COUNTYID', 'OWNNAME1', 'OWNZIP', 'OWNZIP2', 'OWNNAME2', 'OWNCITY', 'OWNSTATE', 'OWNADDRESS', 'LEGAL', 'ADDITIONNU', 'PARHALF', 'PARPD', 'UNITNUM', 'GlobalID', 'last_edite', 'Shape_Leng', 'Shape_Area', 'DEPARTMENT', 'FORM_ACCES', 'FORM_SIGNE', 'FORM_DATE', 'created_da', 'last_edi_1', 'PARPR', 'PARTYPE', 'COUNTY', 'PARSTREET', 'LegalStart', 'ACTIVITY'], inplace = True)\n", "\n", "# Fire dataset\n", "building_fires.drop(columns = ['FID', 'Join_Count', 'TARGET_FID', 'id', 'type', 'lat', 'lon', 'id2'], inplace = True)\n", "building_fires = building_fires.groupby('ADDRESS').max()\n", "building_fires['INCIDENT'] = 1\n", "df = pd.merge(df, building_fires, how = 'left', on='ADDRESS')\n", "df.INCIDENT.fillna(0, inplace = True)\n", "df.date.fillna(pd.to_datetime('1/1/2008'), inplace = True)\n", "\n", "# Foreclosure dataset\n", "foreclosure_df = pd.DataFrame(foreclosures.groupby('ADDRESS').count())\n", "foreclosure_df.rename(columns = { 'NAME' : 'FORECLOSED'}, inplace = True)\n", "foreclosure_df.drop(columns = ['AUCTIONDATE', 'ASSESSEDVALUE', 'YEAR'], inplace = True)\n", "df = pd.merge(df, foreclosure_df, how = 'left', on = 'ADDRESS')\n", "df.FORECLOSED.fillna(0, inplace = True)\n", "\n", "# Rent registry dataset\n", "rent_reg.drop(columns = ['City_1', 'State_1', 'Permit_Num', 'Contact_Ty', 'Last_Name', 'First_Name', 'Middle_Ini', 'Business_N', 'Contact_Pr', 'Business_P', 'Mobile_Pho', 'Home_Phone', 'Email', 'Contact_Ad', 'Issue_Date'], inplace = True)\n", "rent_reg.Units.fillna(value=rent_reg.Units.median(), inplace = True)\n", "rent_reg_grouped = rent_reg.groupby('Address').mean()\n", "rent_df = pd.DataFrame(rent_reg_grouped)\n", "rent_df.rename(columns = { 'Address' : 'ADDRESS', 'Units' : 'RENT_REG_UNITS', 'YEAR' : 'RENT_REG_YEAR'}, inplace = True)\n", "rent_df.rename_axis('ADDRESS', inplace = True)\n", "rent_df['RENT_REG'] = 1\n", "df = pd.merge(df, rent_df, how = 'left', on = 'ADDRESS')\n", "df.RENT_REG.fillna(0, inplace = True)\n", "df.RENT_REG_UNITS.fillna(0, inplace = True)\n", "df.RENT_REG_YEAR.fillna(0, inplace = True)\n", "\n", "# Utility disconnects dataset\n", "utility_disconnects.rename(columns = {'Address' : 'ADDRESS', 'Year' : 'UTILITY_DISCONNECTS'}, inplace = True)\n", "uti_disc_grouped = utility_disconnects.groupby('ADDRESS').count()\n", "uti_disc_df = pd.DataFrame(uti_disc_grouped)\n", "uti_disc_df['ANY_DISCONNECT'] = 1\n", "df = pd.merge(df, uti_disc_df, how = 'left', on = 'ADDRESS')\n", "df.UTILITY_DISCONNECTS.fillna(0, inplace = True)\n", "df.ANY_DISCONNECT.fillna(0, inplace = True)\n", "\n", "# Code cases\n", "case_types = pd.get_dummies(code_cases.CaseType)\n", "code_cases = pd.concat([code_cases, case_types], axis=1)\n", "code_cases.drop(columns = ['Year'], inplace = True)\n", "code_cases_grouped = code_cases.groupby(['ADDRESS']).sum()\n", "code_cases_df = pd.DataFrame(code_cases_grouped)\n", "code_cases_df['TOTAL_VIOLATIONS'] = code_cases_df.sum(axis=1)\n", "code_cases_df['ANY_VIOLATIONS'] = 1\n", "df = pd.merge(df, code_cases_df, how = 'left', on = 'ADDRESS')\n", "df.ANY_VIOLATIONS.fillna(0, inplace = True)\n", "df['Building Service'].fillna(0, inplace = True)\n", "df['Drainage'].fillna(0, inplace = True)\n", "df['Erosion and Sediment Control'].fillna(0, inplace = True)\n", "df['Fire'].fillna(0, inplace = True)\n", "df['Health Nuisance Complaints'].fillna(0, inplace = True)\n", "df['IMPORT / Tree Survey & Stump Removal'].fillna(0, inplace = True)\n", "df['Illicit Discharge'].fillna(0, inplace = True)\n", "df['Landfill'].fillna(0, inplace = True)\n", "df['Manufactured Housing'].fillna(0, inplace = True)\n", "df['Parks and Rec - Dead or Diseased Tree'].fillna(0, inplace = True)\n", "df['Parks and Rec - Tree Complaint'].fillna(0, inplace = True)\n", "df['Property Maintenance'].fillna(0, inplace = True)\n", "df['Rental Registration'].fillna(0, inplace = True)\n", "df['Right of Way '].fillna(0, inplace = True)\n", "df['Sidewalks and Ramps'].fillna(0, inplace = True)\n", "df['Snow'].fillna(0, inplace = True)\n", "df['Special Assessment'].fillna(0, inplace = True)\n", "df['Vegetation'].fillna(0, inplace = True)\n", "df['Waste Water'].fillna(0, inplace = True)\n", "df['Water Purfication'].fillna(0, inplace = True)\n", "df['Zoning'].fillna(0, inplace = True)\n", "df.TOTAL_VIOLATIONS.fillna(0, inplace = True)\n", "\n", "# Crime\n", "crime.rename(columns = {'Offense' : 'CRIME_INCIDENT'}, inplace = True)\n", "crime_grouped = crime.groupby('ADDRESS').count()\n", "crime_df = pd.DataFrame(crime_grouped[['CRIME_INCIDENT']])\n", "crime_df['ANY_CRIME'] = 1\n", "df = pd.merge(df, crime_df, how = 'left', on = 'ADDRESS')\n", "df.CRIME_INCIDENT.fillna(0, inplace = True)\n", "df.ANY_CRIME.fillna(0, inplace = True)\n", "\n", "# Tax assessment\n", "minnehaha_tax.drop(columns=['TAG', 'COUNTYID', 'ACREAGE', 'SQFT', 'FRONTFOOT', 'PARHOUSE',\n", " 'PARHALF', 'PARPR', 'PARSTREET', 'PARTYPE', 'PARPD', 'LOT',\n", " 'BLOCK', 'TRACT', 'SUBDIVNO', 'ADDITION', 'LEGAL', 'COUNTY', 'MRTNSP',\n", " 'MRSCHD', 'SCHOOLDESC', 'MRZON1', 'NUMUNITS', 'MRLYAP', 'MAP_ID'], inplace = True)\n", "lincoln_tax['ADDRESS'] = lincoln_tax.Address\n", "lincoln_tax.drop(columns = ['FID', 'OBJECTID', 'PID', 'Plat', 'SchoolDist', 'Township', 'STR',\n", " 'Address', 'Name', 'Add1', 'Add2', 'Add3', 'Zip', 'Legal1', 'Legal2',\n", " 'Legal3', 'Legal4', 'Class1_1', 'Class2_1','Class3_1', 'Class4_1', 'Class5_1'], inplace = True)\n", "combined_tax = pd.merge(lincoln_tax, minnehaha_tax, how = 'outer', on = 'ADDRESS')\n", "combined_tax.fillna(0, inplace = True)\n", "\n", "combined_tax['LANDVALUE'] = combined_tax.MRLNVC + combined_tax.Value1_1\n", "combined_tax['BUILDVALUE'] = combined_tax.MRBDVC + combined_tax.Value12_1\n", "combined_tax['TOTALVALUE'] = combined_tax[['MRTOTC' , 'Value1_1', 'Value12_1', 'Value13_1', 'Value14_1', 'Value15_1']].sum(axis=1) \n", "\n", "combined_tax.drop(columns=['Value1_1', 'Value12_1', 'Value13_1','Value14_1', 'Value15_1', 'MRLNVC', 'MRBDVC', 'MRTOTC'], inplace = True)\n", "\n", "df = pd.merge(df, combined_tax, how = 'left', on = 'ADDRESS')\n", "df = df[df.ADDRESS != '0']\n", "df = df.dropna(subset=['ADDRESS'])\n", "df.LANDVALUE.fillna(0, inplace = True)\n", "df.BUILDVALUE.fillna(0, inplace = True)\n", "df.TOTALVALUE.fillna(0, inplace = True)\n", "df.set_index('ADDRESS', inplace = True)" ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [], "source": [ "# Moving the incident column to the last position\n", "df1 = df.pop('INCIDENT')\n", "df['INCIDENT'] = df1" ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.0 54127\n", "1.0 1121\n", "Name: INCIDENT, dtype: int64" ] }, "execution_count": 11, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Calculate all addresses with at least one fire incident\n", "df.INCIDENT.value_counts()" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "## Hide the 2018 data\n", "# First create a new column for parcels with incidents in 2018\n", "df['INCIDENT_2018'] = [1 if date > pd.to_datetime('2018') else 0 for date in df['date']] \n", "# Now turn all 2018 fire incidents off\n", "df.loc[(df.date > '2018') & (df.INCIDENT == 1), 'INCIDENT'] = 0" ] }, { "cell_type": "code", "execution_count": 13, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "0.0 54221\n", "1.0 1027\n", "Name: INCIDENT, dtype: int64" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Calculate all addresses with at least one fire incident, excluding 2018 fires\n", "df.INCIDENT.value_counts()" ] }, { "cell_type": "code", "execution_count": 14, "metadata": {}, "outputs": [], "source": [ "# Split predictor and prediction variables\n", "X = df.drop(columns = ['INCIDENT', 'date'])\n", "y = df.INCIDENT" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Apply random forest algorithm" ] }, { "cell_type": "code", "execution_count": 15, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 33] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13 35] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 6 7 13 28 38 39 51 65] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "{'clf__max_depth': 10, 'clf__n_estimators': 100, 'feature_selection__k': 20}\n" ] } ], "source": [ "# Optimizing the algorithm for the best possible results\n", "\n", "# Create pipeline with feature selector and random forest classifier\n", "pipe = Pipeline([\n", " ('feature_selection', SelectKBest(f_classif)),\n", " ('clf', RandomForestClassifier(random_state=2))])\n", "\n", "# Create a parameter grid to test\n", "params = {\n", " 'feature_selection__k':[3, 5, 10, 20, 50],\n", " 'clf__n_estimators':[2, 5, 10, 100],\n", " 'clf__max_depth' : [3, 5, 10, 20, 50]}\n", "\n", "# Initialize the grid search object\n", "grid_search = GridSearchCV(pipe, param_grid=params)\n", "\n", "# Fit it to the data and print the best value combination\n", "print(grid_search.fit(X, y).best_params_)" ] }, { "cell_type": "code", "execution_count": 16, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:113: UserWarning: Features [ 7 13] are constant.\n", " UserWarning)\n", "C:\\Users\\User\\Anaconda3\\lib\\site-packages\\sklearn\\feature_selection\\univariate_selection.py:114: RuntimeWarning: invalid value encountered in true_divide\n", " f = msb / msw\n" ] }, { "data": { "text/plain": [ "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n", " max_depth=10, max_features='auto', max_leaf_nodes=None,\n", " min_impurity_decrease=0.0, min_impurity_split=None,\n", " min_samples_leaf=1, min_samples_split=2,\n", " min_weight_fraction_leaf=0.0, n_estimators=100, n_jobs=1,\n", " oob_score=False, random_state=None, verbose=0,\n", " warm_start=False)" ] }, "execution_count": 16, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Select the best features based on the optimized parameters\n", "sk = SelectKBest(f_classif, k=grid_search.best_params_['feature_selection__k'])\n", "which_selected = sk.fit(X, y).get_support()\n", "X = X[X.columns[which_selected]]\n", "\n", "# Fit the classifier with the optimized parameters\n", "random_forest = RandomForestClassifier(n_estimators=grid_search.best_params_['clf__n_estimators'], max_depth = grid_search.best_params_['clf__max_depth'])\n", "random_forest.fit(X, y)" ] }, { "cell_type": "code", "execution_count": 17, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Correlation heat map with our predictive variables\n", "comb = pd.concat([X, y], axis = 1)\n", "corr = comb.corr()\n", "\n", "plt.figure(figsize = (10,10))\n", "\n", "dropSelf = np.zeros_like(corr)\n", "dropSelf[np.triu_indices_from(dropSelf)] = True\n", "\n", "\n", "ax = sns.heatmap(\n", " corr, \n", " vmin=-1, vmax=1, center=0,\n", " cmap=sns.diverging_palette(20, 220, n=200),\n", " square=True, annot=True, annot_kws={\"size\": 7},\n", " mask=dropSelf\n", ")\n", "\n", "\n", "ax.set_xticklabels(\n", " ax.get_xticklabels(),\n", " rotation=45,\n", " horizontalalignment='right'\n", ")\n", "\n", "plt.savefig('Results - Correlation matrix.pdf', bbox_inches = 'tight', pad_inches = 2.5)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Cross validation and 2018 fire prediction accuracy" ] }, { "cell_type": "code", "execution_count": 18, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "98.1502" ] }, "execution_count": 18, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Cross validation score for Random Forest\n", "Score = cross_val_score(random_forest, X, y, cv=3)\n", "round(np.mean(Score)*100, ndigits=4)" ] }, { "cell_type": "code", "execution_count": 19, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "count 55248.000000\n", "mean 0.018395\n", "std 0.057942\n", "min 0.001134\n", "25% 0.005648\n", "50% 0.006841\n", "75% 0.012729\n", "max 0.989337\n", "Name: PREDICTION, dtype: float64" ] }, "execution_count": 19, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Assign a probability score to each address\n", "df['PREDICTION'] = random_forest.predict_proba(X)[:,1]\n", "df.PREDICTION.describe()" ] }, { "cell_type": "code", "execution_count": 20, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Total 2018 predicted fires (2000 interventions): 42\n", "Total random predicted fires (2000 interventions): 6\n" ] } ], "source": [ "# Compare fires found in 2018 in the 2000 highest risk properties versus randomness\n", "np.random.seed(0)\n", "df['RANDOM'] = np.random.rand(df.shape[0])\n", "df_random = df.nlargest(2000, columns = 'RANDOM')\n", "\n", "rf_results_search = df.nlargest(2000, columns = 'PREDICTION')\n", "print('Total 2018 predicted fires (2000 interventions): ' + str(rf_results_search.INCIDENT_2018.sum()))\n", "print('Total random predicted fires (2000 interventions): ' + str(df_random.INCIDENT_2018.sum()))" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Describe and export the results" ] }, { "cell_type": "code", "execution_count": 21, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 21, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualize how fire risk is distributed among parcel addresses\n", "df.PREDICTION.plot.hist()" ] }, { "cell_type": "code", "execution_count": 22, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Plotting the comparison between randomness and sorting by highest risk with our algorithm\n", "random_results = df_random.INCIDENT_2018.sum()\n", "\n", "rf_results = rf_results_search.INCIDENT_2018.sum()\n", "\n", "df_results = pd.DataFrame({'Model' : ['Random search', 'RF - Highest risk'], 'results' : [random_results, rf_results]})\n", "ax = df_results.plot.bar(x = 'Model', y = 'results', color = ['orange', 'b'])\n", "\n", "plt.title('Results based on 2000 addresses in a year')\n", "plt.ylabel('Fires catched')\n", "ax.set_xticklabels(\n", " ax.get_xticklabels(),\n", " rotation=45,\n", " horizontalalignment='right'\n", ")\n", "\n", "\n", "plt.savefig('Results - Model comparison.pdf', bbox_inches = 'tight', pad_inches = 0.5)" ] }, { "cell_type": "code", "execution_count": 23, "metadata": {}, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Comparing our predictive model to randomness\n", "model_comparison = []\n", "\n", "for searches in range(0, 55000, 100):\n", " prediction = df.nlargest(searches, columns = 'PREDICTION').INCIDENT_2018.sum()\n", " random = df.nlargest(searches, columns = 'RANDOM').INCIDENT_2018.sum()\n", " model_comparison.append([prediction, random])\n", " \n", "model_comparison_df = pd.DataFrame(model_comparison, columns=['PREDICTION', 'RANDOM'])\n", "\n", "model_comparison_df.plot()\n", "plt.xlabel('Properties inspected or intervened')\n", "plt.ylabel('Fire occurance in properties inspected')\n", "plt.xticks([])\n", "\n", "plt.savefig('Model comparison - properties intervened.pdf', bbox_inches = 'tight', pad_inches = 2.5)" ] }, { "cell_type": "code", "execution_count": 24, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nO3dd3gU5fbA8e+hV0VAUUEEBBFUQERAREC9FuyKIhasXMSCinKxi/Veu9hQFP1ZL6hcRayASlFRAQVpItKEAEoRkBYhyfn9cSZmCclmU3ZnNzmf59lndmdmd85ONnt23nnnvKKqOOecc/kpF3YAzjnnkpsnCuecc1F5onDOOReVJwrnnHNReaJwzjkXlScK55xzUXmicDETkQtFZFzYcSQTEdksIk1C2G4jEVERqZDobceDiMwVkW5FeJ5/JhPAE0WKEpGlIrIt+KL6TUReEZEa8dymqr6pqifEcxuRRKSTiHwhIptEZKOIfCAiLRO1/TzimSgifSLnqWoNVV0cp+0dKCLviMja4P3PEpEbRaR8PLZXVEHCalqc11DVg1V1YgHb2SU5JvozWVZ5okhtp6lqDaANcBhwa8jxFElev4pF5EhgHPA+sC/QGPgR+Doev+CT7Ze5iBwAfAcsBw5V1d2Bc4F2QM0S3lZo7z3Z9rvLh6r6LQVvwFLgHxGPHwY+inhcGXgUWAb8DjwPVI1YfgYwE/gTWAScFMzfHXgJWAWsAO4HygfLLgW+Cu4/DzyaK6b3gRuD+/sC/wPWAEuA6yLWuxsYBbwRbL9PHu/vS2BoHvM/AV4L7ncD0oDbgLXBPrkwln0Q8dybgd+A14E9gA+DmNcH9xsE6z8AZALpwGbgmWC+Ak2D+68AzwIfAZuwL/oDIuI5AfgZ2AgMBSbl9d6Ddd+I/HvmsbxRsO1Lgve3Frg9Ynl74BtgQ/C3fAaoFLFcgWuAX4AlwbwnscT0J/A9cHTE+uWD/bwoeG/fA/sBk4PX2hLsl/OC9U/FPl8bgClAq1yf3ZuBWcBfQAUiPs9B7NODOH4HHg/mLwu2tTm4HUnEZzJY52BgPPBH8Nzbwv5fLQ230APwWxH/cDv/YzUAZgNPRiwfAowBamO/QD8A/hMsax98WR2PHVXWBw4Klo0GhgHVgb2AqcCVwbK//ymBLsGXigSP9wC2YQmiXPBFchdQCWgCLAZODNa9G9gBnBmsWzXXe6uGfSkfk8f7vgxYFdzvBmQAj2NJoWvwhdU8hn2Q/dyHgudWBeoAPYLt1wTeAUZHbHsiub7Y2TVR/BHs3wrAm8DIYFnd4Ivv7GDZ9cE+yC9R/AZcFuXv3yjY9otB7K2xL90WwfLDgY7BthoBPwE35Ip7fLBvspPnRcE+qADcFMRQJVj2L+wz1hyQYHt1cu+D4HFbYDXQAUswl2Cf18oRn92ZWKKpGjEv+/P8DdA7uF8D6JjrPVeI2Nal5Hwma2JJ8SagSvC4Q9j/q6XhFnoAfiviH87+sTZjv+4U+ByoFSwT7Asz8tfskeT8chwGPJHHa9YLvmwijzzOByYE9yP/KQX7hdclePxP4IvgfgdgWa7XvhX4v+D+3cDkKO+tQfCeDspj2UnAjuB+N+zLvnrE8reBO2PYB92A7dlfhPnE0QZYH/F4IgUniuERy04G5gf3Lwa+iVgmWKLNL1HsIDjKy2d59pdmg4h5U4Fe+ax/A/BerriPLeAzth5oHdz/GTgjn/VyJ4rngPtyrfMz0DXis3t5Hp/n7EQxGbgHqJvPe84vUZwPzIjn/11ZvXn7YGo7U1U/E5GuwH+xX60bgD2xX8Xfi0j2uoL9ugP7JfdxHq+3P1ARWBXxvHLYF9pOVFVFZCT2zzkZuABrLsl+nX1FZEPEU8pjzUnZdnnNCOuBLGAfYH6uZftgzSx/r6uqWyIe/4od1RS0DwDWqGr63wtFqgFPYMloj2B2TREpr6qZUeKN9FvE/a3YL2KCmP5+z8H+S4vyOuuw91qk7YnIgdiRVjtsP1TAjvIi7fQ3EJGbgD5BrArshn2mwD4zi2KIB+zvf4mI9I+YVyl43Ty3ncsVwL3AfBFZAtyjqh/GsN3CxOgKwU9mlwKqOgn7NftoMGst1gx0sKrWCm67q534BvsnPSCPl1qOHVHUjXjebqp6cD6bHgGcIyL7Y0cR/4t4nSURr1FLVWuq6smRYUd5P1uw5odz81jcEzt6yraHiFSPeNwQWBnDPsgrhpuwppUOqrob1rwGlmCixhyDVdiRkr2gZa8G+a/OZ1gzWFE9hyXZZsF7uY2c95Ht7/cjIkdj5w16Anuoai2seTL7Ofl9ZvKyHHgg19+/mqqOyGvbuanqL6p6Ptb0+RAwKvgbF7T/CxOjKwRPFKXHEOB4EWmjqllY2/UTIrIXgIjUF5ETg3VfAi4TkeNEpFyw7CBVXYX1NHpMRHYLlh0QHLHsQlVnYCd+hwNjVTX7CGIq8KeI3CwiVUWkvIgcIiJHFOL93IL9Kr1ORGqKyB4icj/WfHRPrnXvEZFKwZfdqcA7MeyDvNTEkssGEakNDM61/HfsfEtRfAQcKiJnBj19rgH2jrL+YKCTiDwiInsH8TcVkTdEpFYM26uJnRPZLCIHAVfFsH4G9vesICJ3YUcU2YYD94lIMzGtRKROsCz3fnkR6CciHYJ1q4vIKSISU28tEblIRPYM/obZn6nMILYs8v8bfAjsLSI3iEjl4HPTIZZtuug8UZQSqroGeA1rnwf7dbgQ+FZE/sR+oTYP1p2KnRR+AvvVOAlrLgBrS68EzMOagEYRvQlkBPAPrOkrO5ZM4DSsjX8J9ut+ONajKtb38xVwInbydxXWpHQY0FlVf4lY9bcgzpXYyeN+qprdXJXvPsjHEOzE8FrgW+DTXMufxI6g1ovIU7G+l+D9rMWOkB7GmpVaYj17/spn/UVYUmwEzBWRjdgR23TsvFRBBmLNgZuwL+63Clh/LNajbAG2r9PZuXnocez8zzgsAb2E7Suwc06visgGEempqtOxc1bPYH+bhdi5hFidhL3nzdg+76Wq6aq6Fet99nWwrY6RT1LVTVgHjdOwz8UvwDGF2K7LR3aPFedSTnAl7xuqGq0JJymJSDmse+6Fqjoh7Hici8aPKJxLEBE5UURqiUhlcs4ZfBtyWM4VKG6JQkReFpHVIjInn+UiIk+JyMKgNEHbeMXiXJI4EuuVsxZrHjlTVbeFG5JzBYtb05OIdMH6+b+mqofksfxkoD/W17wDdrGYn3hyzrkkE7cjClWdjF2lmp8zsCSiqvotUEtEYuk37pxzLoHCvOCuPjv3qkgL5q3KvaKI9AX6AlSvXv3wgw46KCEBOudcMps9G7Zvj77O3qxiH35jBllrVXXPomwnzESR++IfyOeCGlV9AXgBoF27djp9+vR4xuWccwBs2gR/5dmBOXyZmbDPPjBgANx4Yx4rqIIIVcaNocrkcdR89dlfi7qtMBNFGnbJfbYGWF9455wL3fffQ/v2kJUVdiTRHXIINIjsIL5+PQwcCE2awO23w+Wn2+3VZ4u8jTATxRjg2qBeUAdgY3BlsHPOhW7RIksSt98Oe0e7hj5ElStDz54RM957D66+GtasgTvuKLHtxC1RiMgIrEJn3aD42WCs4Byq+jxWlO5k7KrNrdiVws45lzD33w9jx+a97PffbfrPf8L+++e9TtL4/Xfo3x/eeQfatIGPPoK2JXfFQdwSRVDUK9ry7IFTnHMuFM88AxUrwoEH7rpsv/2gQweoXz/xcRXa8uWWHB54AP71L3tTJcjLjDvnUtrmzfDqq5CeXvC6kVTth/g998Bdd8Untrj69Vf44AO49lpo1w6WLYM6dQp+XhF4onDOpbSPP7bvyqIQgSMKU9M4GWRlwXPPwS232OMePaz7U5ySBHiicM7FWVYWLF1qv+DjYcUKm06fnncTUjTly0O1aiUfU9z8/DP06QNffQUnngjDhlmSiDNPFM65uBo82E4ax1vdulAzphEvUtTWrdC5s11A8corcPHFdkiUAJ4onHNxtWaNTV97LX7bqFs3BXomFdWCBdCsmR36vP669WpKcH9dTxTOuWLbvh1OPRVW5XEl1IoVUK8e9O6d+LhSWno63HcfPPSQHUFcdBGcdFIooXiicM4V28qVMH48HH74rr/sDzwQOnUKJ66U9fXXcMUVdk7issvglFNCDccThXNuJxMnwuefF+45a9fadNCgXFcKu8K77z47sdOwoV0NeMIJYUfkicI5t7NBg2DaNChXyEEIatSA5tFGJHfRBUX8aNPGrrJ+4AHbqUnAE4VzZUx6Osydm//ydeusa/6oUYmLqUz74w8rAdu0Kdx5J5x2mt2SiCcK58qYm26CoUOjr3PssYmJpcwbNQquucaSxZ13hh1NvjxROFfGbNhg0zFj8l/HTz7H2apVdjn5u+9aD4Bx46B167CjypcnCueS3NVXw/vv21XEJWHtWmvlSLLWjbJl5Uo7Uf3QQzbqUIXk/ipO7uicc4wfb98rl5VgIf5u3UrutVyMli61In79+9tRxPLlsMceYUcVE08UzhWBKjz4YN4XmJW0NWugVy94+eX4b8vFQWYmPPss3HabdSU791y7sjpFkgR4onCuSFassP/7atVslLF4KlcuBSucOvPTT1bEb8oUu6p62LDkHS4vCk8UzhXBjBk2HTbMKis4t4utW6FLFyuf+9pr9kFJUBG/kuaJwrlC2rYNTj/d7sdxCACXqubPtysPq1WDN9+03kz16oUdVbEU8tpL59z69Ta9/HIbEsA5wH5B3HwzHHywJQiw8hspniTAjyhcipo8Gc46y/43E92zMCvLpscdV/gyF66UmjzZzkX88otNTz017IhKlCcKl5J++skuZu3VKyEDfO2iShXo3j3x23VJ6J574O67oXFj+Owz+wVRyniicCnj7bfhww/t/oIFNn388XAShXN/F/Fr185qNd13H1SvHnZUceGJwqWMxx+H776zH24ARx3lJ5NdCNautcTQrBncdZeNFRHyeBHx5i2sLmX8/LOdPF682G5ffQWVKoUdlSszVO2wtmVLGDmyTJ2g8iMKlxK2bLFidqtXhx2JK5NWrswputWunZ2LaNUq7KgSpuykRJdSVHe+bd9u8889N9y4XBn122/wxRfwyCPwzTdlKkmAJwqXhAYNsqP6yFvt2rYsSQb8cmXB4sUwZIjdb9sWli2DgQOTvtJrPJS9d+yS3vz5Nr377p3nV6wI552X8HBcWZOZCU89Bbffbh+6Xr2sPlOtWmFHFhpPFC7uXn7ZmnZjNXUqHHaYjS/vXELNnQtXXGHd6045BZ5/PiWL+JU0TxQu7p57DqZPtzHjY7HPPnbVtXMJtXUrdO1q10b89792JJGiRfxKmicKVyQffRR7D6S1a+0q5o8/jm9MzhXJvHnQooUV8Rs50or47bln2FElFU8UrtBWrSp8KRsfUc0lna1brX3z8cfhlVegd2/4xz/CjiopeaJwhbJ9e86RxN13w6WXxva8+vXjFZFzRTBxIvzzn7BwIVx5ZU7deJcnTxQuZllZVj5j5Up73LQp7L9/uDE5V2iDB8O998IBB9i1EcccE3ZESc8ThYvZpk2WJM46y8rs+48wl1Kyi/i1bw833WTJolq1sKNKCXG94E5EThKRn0VkoYjcksfy3UXkAxH5UUTmishl8YzHFU9amk27d4d+/aBmzXDjcS4ma9bABRdYYgDr9vroo54kCiFuiUJEygPPAt2BlsD5ItIy12rXAPNUtTXQDXhMRLzMW5JatcqmfnW0Swmq1s21RQsYNcorSBZDPJue2gMLVXUxgIiMBM4A5kWso0BNERGgBvAHkBHHmFwMpk2Dr7/edf7cuTY98MDExuNcoaWlwVVX2QAmHTrASy/ZEKWuSOKZKOoDyyMepwEdcq3zDDAGWAnUBM5T1azcLyQifYG+AA0bNoxLsC5H//52YWpeqlaFBg0SG49zhbZmjQ1P+vjjcN11UL582BGltHgmirwuadRcj08EZgLHAgcA40XkS1X9c6cnqb4AvADQrl273K/hSsC6dXayGmDzZujcGT74YNf1qlSxm3NJZ+FC+9AOGGA1YJYvh912CzuqUiGeiSIN2C/icQPsyCHSZcCDqqrAQhFZAhwETI1jXC6XlSuhYUOrhZatR48yXQPNpZKMDKvyeuedULmynbiuV8+TRAmKZ6KYBjQTkcbACqAXcEGudZYBxwFfikg9oDmwOI4xuTz88YcliVNOgXPOsXldu4Ybk3MxmT3bivhNm2b9tYcOtSThSlTcEoWqZojItcBYoDzwsqrOFZF+wfLngfuAV0RkNtZUdbOqro1XTG5nTzxhlQvS0+3xpZfmJArnkt7WrXaxXLlyVqOpZ08v4hcncb3gTlU/Bj7ONe/5iPsrgRPiGYPL3wcfwKxZcMYZNi5Lp05hR+RcDObMsR5M1arBW29ZEb+6dcOOqlTzK7NLqQULYMQI60qenyVL4OijYfToxMXlXJFt2WLnIYYMgVdftSJ+xx0XdlRlgieKUuqZZ+Dppwte78QT4x+Lc8X2+edWxG/JErj6ajsMdgnjiaKUWbnSqrv+9psdja9ZE3ZEzhXTnXfC/fdDs2YwaRJ06RJ2RGWOJ4pSJD3dCmJmn5z2axNdSsvKshPVnTrBoEFW175q1bCjKpM8UZQiW7dakrjiChtYyEttuJS0erVdTd28Odxzj1Wh7N497KjKtLhWj3WJNX++TVu1gjPPhJa5SzA6l8xU4Y03rIjfe+95ddck4omiFFmxwqZNmoQbh3OFtny5HQb37m1HEjNmwM03hx2VC3jTUynw11/w73/D9On2uHHjcONxrtDWrbOSxU8+Cddc40X8kownilLghx9yxmRp1Aj23TfUcJyLzYIFMGYMDBwIbdrYUYWPhpWUPFEkoZ9+svN5sZozx6Zjx9oQpc4ltYwMeOwxG7u6alVrbqpXz5NEEvNEkWTWr4dDDrGegYXl/2cu6f34I1x+uR0Gn3UWPPusF/FLAZ4oksy2bZYkBgyA006L/Xk1akC7dvGLy7li27rVSm5UqGBDk/boEXZELkaeKBLs3/+2W8WKeS/PPpJo0cIKYzqX8mbNgkMPte6u77xjRfxq1w47KlcInigSbNo0q2123XX5r1OpUuGOJpxLSps3w+23W9GxV16Biy/2Xz8pyhNFAk2ebJVaDz3UegE6V2qNHw99+8LSpXDttXY+wqWsmC64E5FKItI03sGUdsOG2fQf/wg3Dufi6vbbrftd5crw5Zd2ROE9LVJagYlCRE4BZgPjg8dtROS9eAdWmmzaZIMELV8OTZvC44+HHZFzcZB9gq1zZ7j1Vpg50+67lBfLEcW9QAdgA4CqzgT86KIQHnvMhvP98kuoVSvsaJwrYb/9ZmPo3n23Pe7e3XpsVKkSaliu5MSSKHao6oZc86KMm+Zy27rVpj/8AOPGhRuLcyVG1U5St2wJH34Iu+0WdkQuTmI5mf2TiPQEyolIY+B64Nv4hpX6Wre2XoHZatSAww4LLx7nStSvv9rJ6nHjrHlp+HAr5udKpVgSxbXAXUAW8C4wFrg1nkGVBrNmWV2z22+3xwcfHG48zpWoDRusr/czz8BVV9kAQ67UiiVRnKiqNwN/1/wVkbOxpOEirF4NN96Y09R0yy027opzpcLPP1sRv3/9yw6Zly2zQ2VX6sXyM+COPObdXtKBlAbffgtvvgmzZ1sxzK5dw47IuRKwYwf85z+WHB58MKdipSeJMiPfIwoRORE4CagvIpEdOnfDmqFcLtkDB731FrRtG24szpWIGTNsbN0ZM6xn0zPPwF57hR2VS7BoTU+rgTlAOjA3Yv4m4JZ4BpWqPvvMpnXqhBuHcyVi61Y4/ngrTPa//8HZZ4cdkQtJvolCVWcAM0TkTVVNT2BMKSkry/6fateG/fcPOxrnimHGDGs7rVbNqry2bg177BF2VC5EsZyjqC8iI0VklogsyL7FPbIU06OHNTnVrRt2JM4V0aZNVpepbVt4/XWb162bJwkXU6J4Bfg/QIDuwNvAyDjGlJKWLrXrjl58MexInCuCTz+1EbOGDoXrr/dmJreTWBJFNVUdC6Cqi1T1DqDM1woeN87GYDn2WLv98ovVcerSJezInCukW2+1shvVq8PXX8OQId6jye0kluso/hIRARaJSD9gBVDmuz2MGQOTJkGnTva4bVvrFOJcysjMtKtCu3WzUefuuMMqvjqXSyyJYgBQA7gOeADYHbg8nkElu48+sqF/a9WyMSacSymrVsE111i5gPvugxNPtJtz+SgwUajqd8HdTUBvABFpEM+gkllmJpx6qt332k0upWQX8bvxRkhP9xLgLmZRz1GIyBEicqaI1A0eHywir+FFAbnhBvjuu4LXcy4pLF1qgwldfrkNsfjjj5YwnItBvolCRP4DvAlcCHwqIrcDE4AfgQMTE17yql3brptwLiVs3Gh17ocOhYkT4cAy/y/sCiFa09MZQGtV3SYitYGVweOfY31xETkJeBIoDwxX1QfzWKcbMASoCKxVVa+Q5FxJmDfPel3ccktOEb/q1cOOyqWgaE1P6aq6DUBV/wDmFzJJlAeexa69aAmcLyItc61TCxgKnK6qBwPnFjL+hFq3zrrEOpfUtm+H+++3k2iPPppTxM+ThCuiaEcUTUQku5S4AI0iHqOqBV2R0x5YqKqLAURkJHaUMi9inQuAd1V1WfCaqwsZf0LNnWtdYjt3zjmh7VxSmT7divjNmgW9esGTT3oRP1ds0RJFj1yPnynka9cHlkc8TsPG3o50IFBRRCYCNYEnVfW13C8kIn2BvgANGzYsZBglZ8cOmz7wgPd4ckloyxbr5lqlCrz/vg3U7lwJiFYU8PNivrbk9bJ5bP9w4DigKvCNiHyrqjvVklLVF4AXANq1axfaeN0ZGTb1k9guqfzwgxXxq14d3nsPWrWyi3ycKyHxHL8wDdgv4nED7IR47nU+VdUtqroWmAy0jmNMRaIKCxbAwoX22BOFSwp//glXXw2HHw5vvGHzunTxJOFKXCxXZhfVNKCZiDTGyn70ws5JRHofeEZEKgCVsKapJ+IYU5GMGQNnnpnzuGbN8GJxDoCPP4Yrr4SVK+16iB65W4qdKzkxJwoRqayqf8W6vqpmiMi1wFise+zLqjo3qBeFqj6vqj+JyKfALGzUvOGqOqdwbyH+1q+36dCh0Ly5d0F3Ibv5Znj4YStXPGoUdMh96s+5klVgohCR9sBLWI2nhiLSGuijqv0Leq6qfgx8nGve87kePwI8UpigEyEjwy5kXbIENm+2ed27Q6NGoYblyipVGx2rfHnro12lCtx2mxfxcwkRyxHFU8CpwGgAVf1RREp9mfHNm2HCBLt/0UXWwzDEDleuLFuxws5FHHqoXR9xwgl2cy5BYkkU5VT1V6s0/rfMOMUTqkWLYNgwK/yXHgz++sQTVtfJuYRTheHDYeBAu4jumFL/+8wlqVgSxfKg+UmDq637A6VyKNQ33oBHHskZs2WPPawZ2LmEW7LELpybMMHGi3jxRRsZy7kQxJIorsKanxoCvwOfBfNKlWXLrMAm2NDBzoVq82a7unrYMOjTB8rFsye7c9HFkigyVLVX3CMJ2RFHWEkcHwHShWbOHOuLfdttdj5i2TKoVi3sqJyL6YK7aSLysYhcIiKl9gqCdeusLMf334cdiStztm+He+6x8XSfeCKniJ8nCZckCkwUqnoAcD9WamO2iIwWkVJzhLF9uyWIzEw4/ni/RsIl2LRpdmX13XfDuedaaXAv4ueSTEwNn6o6RVWvA9oCf2IDGpUKmzbBzJnQtStccknY0bgyZcsWOOkku6JzzBh4803Yc8+wo3JuF7FccFcDKw/eC2iBld3oFOe4EmLcOHgtqFXbo4f3cHIJMn26NTNVr25VXg89FHbfPeyonMtXLEcUc4COwMOq2lRVb1LVUjFa9JNPwltv2dXWbdqEHY0r9TZutPpMRxyRU8Svc2dPEi7pxdLrqYmqZsU9kgTascO6p//+uyWIadPCjsiVeh98AP36wW+/2QV055wTdkTOxSzfRCEij6nqTcD/RGSXMSBiGOEuaf3vf3D++XbfhzZ1cfevf9mQpIceCqNH2xGFcykk2hHFW8G0sCPbJb2tW236wQd25O9ciVO1rnQVKlhdpt12s6qvlSqFHZlzhRZthLupwd0WqrpTsgjKhxd3BLxQTJlilRHAmp18jBdX4tLS4KqrbKS5Bx6wftfHHx92VM4VWSwnsy/PY94VJR1IoixaZNPrr4f69cONxZUyWVlWcqNlS/jiC9h777Ajcq5ERDtHcR7WJbaxiLwbsagmsCHegcXDsGFWjBOgf3+QvEb1dq4oFi+Gyy+HSZPsxNcLL0CTJmFH5VyJiHaOYiqwDhvr+tmI+ZuAGfEMKl6GDLHS/t26wb77hh2NK1W2bLGrqocPt4Thv0JcKRLtHMUSYAlWLTbl/fgjzJ8PPXvatRPOFdvs2XbB3B13WI+mX3+FqlXDjsq5EpfvOQoRmRRM14vIHxG39SLyR+JCLBk9e9rUKyS4YvvrL7jrLru6+qmncor4eZJwpVS0pqfs4bTqJiKQeFu3Do46ypqfnCuyb7+1bnPz5kHv3lbttU6dsKNyLq6iNT1lX429H7BSVbeLSGegFfAGVhwwJdxzjyWKFi2sW7tzRbJlC5xyitVo+vhj6N497IicS4hYuseOxoZBPQB4DSsM+N+4RlXCFi+26c03hxuHS1HffWddX6tXt6s05871JOHKlFgSRZaq7gDOBoaoan8gZa5AGD/eKsQ2bOhDDrtC2rDBhiHt2DGniF+nTlCz1I7f5VyeYhoKVUTOBXoDZwbzKsYvpJL13ns29bEmXKGMHg1XX20nqm++2QYVcq6MivXK7GOwMuOLRaQxMCK+YZWcjAyoVw/uvTfsSFzKuPFGOOssG2nuu+/gwQe9R5Mr0wo8olDVOSJyHdBURA4CFqrqA/EPrWRkZkL58mFH4ZJeZBG/k0+2nkyDBkHFlDl4di5uYhnh7mjgdWAFIMDeItJbVb+Od3DFlZ5uQ516TycX1bJlNlbEYYdZEb9//MNuzjkgtqanJ4CTVfUoVe0EnAI8GfEqKYwAABq3SURBVN+wSkbnzvDOO95q4PKRlQVDh8LBB1uNJq/r4lyeYkkUlVR1XvYDVf0JSImi+qtWWbJ4/fWwI3FJZ+FCK/p1zTVw5JHW5fWaa8KOyrmkFEui+EFEholI5+D2HClQFHDNGli5Epo39wHFXB7S02HBAvi//4OxY23gdOdcnmJpve8HXAcMws5RTAaejmdQJWHOHJs2bBhuHC6JzJxpRfwGD4ZDDoGlS6FKlbCjci7pRU0UInIocADwnqo+nJiQSkZmpk2PPTbcOFwSSE+H++6Dhx6CunVt9Lm99vIk4VyMolWPvQ0r33EhMF5E8hrpLmllZNjUu8aWcVOmWG+mf/8bLrrIivnttVfYUTmXUqIdUVwItFLVLSKyJ/Ax8HJiwiq+7CMK7xpbhm3ZAqedBjVqwKefwoknhh2Rcykp2tfoX6q6BUBV14hILCe+k0b2EYUnijLom2+gQwcr4vfhh3Y+wuszOVdk0b78m4jIu8HtPeCAiMfvRnne30TkJBH5WUQWisgtUdY7QkQyReScwr6B/HjTUxm0fr0NQ9qpU06f6COP9CThXDFF+73dI9fjZwrzwiJSHhtr+3ggDZgmImMir8mIWO8hYGxhXr8g779vUz+iKCPefdeug1izBm69Fc47L+yInCs1og1c9HkxX7s9VhdqMYCIjATOAOblWq8/8D+gRK92yD6iaNasJF/VJaUBA2zowjZtbEChww4LOyLnSpV4/t6uDyyPeJwGdIhcQUTqA2cBxxIlUYhIX6AvQMMYL4z44gs44ACv6VZqRRbxO/VU68k0cKD/wZ2Lg3ieoJY85mmux0OAm1U1M9oLqeoLqtpOVdvtueeeMW18wwZrhXCl0NKlcNJJcOed9vi446y5yZOEc3ERc6IQkcqFfO00bLztbA2AlbnWaQeMFJGlwDnAUBE5k2JQtQtwK1SAyy4rziu5pJOVBU8/bb2YpkyB/fcPOyLnyoQCE4WItBeR2cAvwePWIhJLCY9pQDMRaSwilYBewJjIFVS1sao2UtVGwCjgalUdXdg3EWnSJGui3rLFus+7UuKXX6BLF7juOjj6aKvR0q9f2FE5VybEco7iKeBU7CptVPVHETmmoCepaoaIXIv1ZioPvKyqc0WkX7D8+aKHnb8//7Tp0KHQu3c8tuBCsX07LFpkA6BfdBFIXi2bzrl4iCVRlFPVX2Xnf8yo5xSyqerH2BXdkfPyTBCqemksr1mQK6+0aceOfkSR8mbMsH7Od99tY0YsXQqVC9sC6pwrrljOUSwXkfaAikh5EbkBWBDnuIps40abHnJIuHG4YkhPt5PTRxwBw4bl9ErwJOFcKGJJFFcBNwINgd+BjsG8pLNxI2zbBjfc4B1gUtZXX0Hr1vDgg3DxxVbEL8aebs65+Ciw6UlVV2MnopPelCk23W23cONwRbR5M5xxhv0Bx42D448POyLnHDEkChF5kV2vf0BV+8YlomJYGXS+PeWUcONwhfTVV1afqUYN+Ogjazf0E0zOJY1Ymp4+Az4Pbl8DewF/xTOooho61Ka77x5uHC5G69ZZ89LRR+cU8fNeCM4lnViant6KfCwirwPj4xZRMdSsCfvtZ+NkuySmCqNGwbXXwh9/2BXWvVKiddO5MqkoJTwaA0l5SeykST5GdkoYMAB69rSsPn063Huv92hyLonFco5iPTnnKMoBfwD5ji0Rplq1/DqspKVqJX0rVoTTT4d994Ubb/Q68M6lgKj/pWJX2bUGVgSzslR1lxPbyeCpp6wQYKtWYUfidrFkCfTtC4cfbt1ejz3Wbs65lBC16SlICu+pamZwS8okAfDAAzbt1i3UMFykzEx48knrxfTdd9CkSdgROeeKIJZzFFNFpG3cIymm1avhn/+Ec88NOxIHwIIF1pvphhuga1eYO9eOKpxzKSffpicRqaCqGUBn4J8isgjYgo0zoaqaNMlj82abrl4dbhwuQkYG/PorvPEGXHCBnzxyLoVFO0cxFWgLFGt8iETYvt2mXbuGG0eZN326FfG77z5o2RIWL/beTM6VAtGangRAVRfldUtQfDHJyrKpd6AJybZtMGgQdOgAL7/sRfycK2WifbXuKSI35rdQVR+PQzxFkn2K3Vs3QjBpEvTpAwsX2kmihx+2fsrOuVIjWqIoD9Qg77Gvk0p2jSdPFAm2eTOcfbYlhs8/9y6vzpVS0RLFKlW9N2GRFMOSJTbdY49w4ygzvvwSjjrKajJ98okNKlS9ethROefipMBzFKngiy9s2qJFuHGUemvX2jCkXbrkFPFr396ThHOlXLQjiuMSFkUxPf20TX18mzhRhbffhv79Yf16GDzYi/g5V4bkmyhU9Y9EBlJc/ftDgwZhR1FKXX+9ZeMjjrBzEYceGnZEzrkESvkOpXPn2rRSpXDjKHVUYccO27FnnQX7729XWZcvH3ZkzrkEK0qZ8aTy++82PfLIcOMoVRYtguOOgzvusMfHHAM33eRJwrkyKuUTxTff2NTPT5SAzEx4/HFrWvr+ex8ByjkHlIKmpz//tKk3mxfT/PlwySUwdSqcdho89xzUrx92VM65JJDyiWLhQihXzq+hKLasLLtyccQIOO88v3rROfe3lE8U6ek5tZ5cIU2dakX8HnjAivgtWuS9Apxzu0j5cxSZmfYd5wph61YYONB6ALz6ak4RP08Szrk8pHyimD7dipe6GE2YYCd0HnvMivjNnes9AZxzUaV809M++3g165ht3mxDANaqZQnDx411zsUgpY8ofvkF5syB/fYLO5IkN3GincjJLuI3a5YnCedczFI6UcyebdOOHcONI2mtWQPnn28XzL3xhs074gioVi3cuJxzKSWlm57S0mzavXu4cSQdVevmet11sGmTDU3qRfycc0WU0onij6Bs4V57hRtH0unfH5591g61XnrJu4U554olpRNF9hCo9eqFG0dSyMqCjAzr4nrOOdC0qSUMr8/knCumuJ6jEJGTRORnEVkoIrfksfxCEZkV3KaISOvCvP6OHVChgl9EzC+/2DCkt99uj7t180qvzrkSE7dEISLlgWeB7kBL4HwRyd0GsgToqqqtgPuAFwqzjR07oGLFkog2RWVkwKOPQqtWMHOmD/HnnIuLeDY9tQcWqupiABEZCZwBzMteQVWnRKz/LVCooYcmTsxpfipzfvoJLr7Yrjg84wwYOhT23TfsqJxzpVA8m57qA8sjHqcF8/JzBfBJXgtEpK+ITBeR6Wuyy00AtWvbj+oy6/ff4a234L33PEk45+ImnokirzMHef7+F5FjsERxc17LVfUFVW2nqu32jCg3oQrt2pVEqCni22/h1lvtfosWVsSvZ08/SeOci6t4Joo0IPKa6QbAytwriUgrYDhwhqquK8wGVK3EeKm3ZQsMGACdOsGbb+YU8SvTJ2icc4kSz6/ZaUAzEWksIpWAXsCYyBVEpCHwLtBbVRcUdgOrV5eBH9OffQaHHAJDhsDVV3sRP+dcwsXtZLaqZojItcBYoDzwsqrOFZF+wfLngbuAOsBQsW/8DFWNuTHpt9/gr79KPvaksXmzXVFduzZMngxHHx12RM65MiiuF9yp6sfAx7nmPR9xvw/Qp6ivv9tu0KRJ0eNLWl98AV27WhG/sWPtyuqqVcOOyjlXRqV8C3+pGgL199/t5PRxx+UU8Tv8cE8SzrlQpXSiyMoqJSezVeH11+3IIXto0gsuCDsq55wDUrzWU6lJFNdcA889Z0OTvvSSX2HtnEsqnijCkpVlNUgqV4bzzrPkcPXVXp/JOZd0UvVrFoBly1I0Ufz8s52szi7i17WrV3p1ziWtVPyaBWDFCptmj0mREnbsgAcfhNatbQzXQw8NOyLnnCtQyjY9bdtm0zPOCDeOmM2dC717w4wZcPbZNrDQ3nuHHZVzzhUoZRNFtgqp8g7Kl7fDn1GjoEePsKNxzrmYpWzTU0qUF58yBW4O6hwedBAsXOhJwjmXclI2UWRLylpPmzfDdddB585WBnztWpufMoc/zjmXI+UTRdIZN86K+D3zDFx7rZ20rls37Kicc67IUvYn7q+/hh1BHjZvhgsvhDp14Msv4aijwo7IOeeKLWWPKF4IRtdOio5D48dDZqYV8Rs3zsav9iThnCslUjZRiFiLznHHhRjEqlV2cvqEE2xAIYDDDoMqVUIMyjnnSlbKJoqvvgqxqKoqvPKKFfH76CO7iM6L+DnnSqmUPUexzz6waVNIG7/qKhg2zHo1DR8OzZuHFIhzyW3Hjh2kpaWRnp4edihlRpUqVWjQoAEVS3Co5JRNFJmZdmlCwkQW8bvgAmjVCvr1S9FiU84lRlpaGjVr1qRRo0ZIUvZlL11UlXXr1pGWlkbjxo1L7HVT9lsuIyOBlyX89JMNQ3rbbfa4Sxer9OpJwrmo0tPTqVOnjieJBBER6tSpU+JHcCn7TZeQRLFjB/z739CmDcyfbyeqnXOF4kkiseKxv1O26SnuiWLuXLjoIuvqeu658PTTUK9eHDfonHPJKWWPKNati3OiqFABNm6Ed9+Ft9/2JOFcCnvvvfcQEebPn//3vIkTJ3LqqafutN6ll17KqFGjADsRf8stt9CsWTMOOeQQ2rdvzyeffFLsWP7zn//QtGlTmjdvztixY/NcZ+bMmXTs2JE2bdrQrl07pk6dutPyZcuWUaNGDR599NFixxOLlEwUy5fD+vVQ4h0pvvwSBg60+82bw4IFcNZZJbwR51yijRgxgs6dOzNy5MiYn3PnnXeyatUq5syZw5w5c/jggw/YVMyulvPmzWPkyJHMnTuXTz/9lKuvvprMzMxd1hs0aBCDBw9m5syZ3HvvvQwaNGin5QMGDKB79+7FiqUwUrLpKXuwom7dSugFN22CW26BoUOhcWO7X7euF/FzrgTdcIO15JakNm1gyJDo62zevJmvv/6aCRMmcPrpp3P33XcX+Lpbt27lxRdfZMmSJVSuXBmAevXq0bNnz2LF+/7779OrVy8qV65M48aNadq0KVOnTuXII4/caT0R4c8//wRg48aN7Lvvvn8vGz16NE2aNKF69erFiqUwUvKbcMMGm5ZIa9Ann8CVV0Jamn2S778fEvgHcM7F1+jRoznppJM48MADqV27Nj/88ANt27aN+pyFCxfSsGFDdttttwJff8CAAUyYMGGX+b169eKWW27Zad6KFSvo2LHj348bNGjAiuzhOiMMGTKEE088kYEDB5KVlcWUKVMA2LJlCw899BDjx49PWLMTpGiiyB6LothjUmzaBBdfDHvtZWNHRPwBnXMlq6Bf/vEyYsQIbrjhBsC+vEeMGEHbtm3z7R1U2F5DTzzxRMzrah5fWnlt77nnnuOJJ56gR48evP3221xxxRV89tlnDB48mAEDBlCjRo1CxVhcKZkospv09tyzCE9WhbFj4fjjoWZN+Owzu3IvOLx0zpUe69at44svvmDOnDmICJmZmYgIDz/8MHXq1GH9+vU7rf/HH39Qt25dmjZtyrJly9i0aRM1a9aMuo3CHFE0aNCA5cuX//04LS1tp2albK+++ipPPvkkAOeeey59+vQB4LvvvmPUqFEMGjSIDRs2UK5cOapUqcK1114b2w4pKlVNqdvhhx+uY8eqgupXX2nhrFypeuaZ9uRXXy3kk51zhTVv3rxQt//8889r3759d5rXpUsXnTx5sqanp2ujRo3+jnHp0qXasGFD3bBhg6qq/utf/9JLL71U//rrL1VVXblypb7++uvFimfOnDnaqlUrTU9P18WLF2vjxo01IyNjl/UOOuggnTBhgqqqfvbZZ9q2bdtd1hk8eLA+8sgjeW4nr/0OTNcifu+m9BFF+fIxPkEV/u//4MYb4a+/4OGHvYifc2XAiBEjdvlV36NHD/773/9y9NFH88Ybb3DZZZeRnp5OxYoVGT58OLvvvjsA999/P3fccQctW7akSpUqVK9enXvvvbdY8Rx88MH07NmTli1bUqFCBZ599lnKB19kffr0oV+/frRr144XX3yR66+/noyMDKpUqcIL2eMqhEQ0JQafztGuXTu9557pnHoqTJ0KRxwRw5OuvNIGsOjSxYr4NWsW9zidc/DTTz/RokWLsMMoc/La7yLyvaq2K8rrpfQRRdRSS5mZVoKjShW7wvqww6BvX6/P5JxzhZSS35oFNj3NnWsjzGUX8Tv6aK/06pxzRZSS35xZWTbdJVFs3w733WdHDwsXxtgu5ZyLp1Rr3k518djfKdn0tGyZTXc6QJg9Gy680Ka9esFTTxWx/6xzrqRUqVKFdevWeanxBNFgPIoqJTwcc0omiu3bbVq7dsTMSpVg61Z4/304/fRQ4nLO7axBgwakpaWxZs2asEMpM7JHuCtJKZkoNm60aa0fJ8GjY+Cxx6yI388/F6LPrHMu3ipWrFiiI625cMT1HIWInCQiP4vIQhG5JY/lIiJPBctniUj0AiyBmZP/ZChXUbV7Nxg9GtautQWeJJxzrsTFLVGISHngWaA70BI4X0Ra5lqtO9AsuPUFnivwhTdu5LUfDqYvL9gFdLNnW6VX55xzcRHPI4r2wEJVXayq24GRwBm51jkDeC24wvxboJaI7BPtRTMWLuW3bbtzSdMp1uRUrVp8onfOOQfE9xxFfWB5xOM0oEMM69QHVkWuJCJ9sSMOgL8OZe4cFnbkTe9EURdYG3YQScL3RQ7fFzl8X+RoXtQnxjNR5PU1nruDbyzroKovAC8AiMj0ol6GXtr4vsjh+yKH74scvi9yiMj0oj43nk1PacB+EY8bACuLsI5zzrkQxTNRTAOaiUhjEakE9ALG5FpnDHBx0PupI7BRVVflfiHnnHPhiVvTk6pmiMi1wFigPPCyqs4VkX7B8ueBj4GTgYXAVuCyGF463Hq7ycX3RQ7fFzl8X+TwfZGjyPsi5cqMO+ecS6yULAronHMucTxROOeciyppE0W8yn+kohj2xYXBPpglIlNEpHUYcSZCQfsiYr0jRCRTRM5JZHyJFMu+EJFuIjJTROaKyKREx5goMfyP7C4iH4jIj8G+iOV8aMoRkZdFZLWIzMlnedG+N4s62HY8b9jJ70VAE6AS8CPQMtc6JwOfYNdidAS+CzvuEPdFJ2CP4H73srwvItb7AusscU7YcYf4uagFzAMaBo/3CjvuEPfFbcBDwf09gT+ASmHHHod90QVoC8zJZ3mRvjeT9YgiLuU/UlSB+0JVp6jq+uDht9j1KKVRLJ8LgP7A/4DViQwuwWLZFxcA76rqMgBVLa37I5Z9oUBNsUExamCJIiOxYcafqk7G3lt+ivS9mayJIr/SHoVdpzQo7Pu8AvvFUBoVuC9EpD5wFvB8AuMKQyyfiwOBPURkooh8LyIXJyy6xIplXzwDtMAu6J0NXK+qWYkJL6kU6XszWcejKLHyH6VAzO9TRI7BEkXnuEYUnlj2xRDgZlXNLOUjqsWyLyoAhwPHAVWBb0TkW1VdEO/gEiyWfXEiMBM4FjgAGC8iX6rqn/EOLskU6XszWROFl//IEdP7FJFWwHCgu6quS1BsiRbLvmgHjAySRF3gZBHJUNXRiQkxYWL9H1mrqluALSIyGWgNlLZEEcu+uAx4UK2hfqGILAEOAqYmJsSkUaTvzWRtevLyHzkK3Bci0hB4F+hdCn8tRipwX6hqY1VtpKqNgFHA1aUwSUBs/yPvA0eLSAURqYZVb/4pwXEmQiz7Yhl2ZIWI1MMqqS5OaJTJoUjfm0l5RKHxK/+RcmLcF3cBdYChwS/pDC2FFTNj3BdlQiz7QlV/EpFPgVlAFjBcVfPsNpnKYvxc3Ae8IiKzseaXm1W11JUfF5ERQDegroikAYOBilC8700v4eGccy6qZG16cs45lyQ8UTjnnIvKE4VzzrmoPFE455yLyhOFc865qDxRuKQTVH2dGXFrFGXdRvlVyizkNicG1Ud/FJGvRaR5EV6jX3aZDBG5VET2jVg2XERalnCc00SkTQzPuSG4jsK5IvFE4ZLRNlVtE3FbmqDtXqiqrYFXgUcK++Tg2oXXgoeXAvtGLOujqvNKJMqcOIcSW5w3AJ4oXJF5onApIThy+FJEfghunfJY52ARmRochcwSkWbB/Isi5g8TkfIFbG4y0DR47nEiMkNEZge1/isH8x8UkXnBdh4N5t0tIgPFxsBoB7wZbLNqcCTQTkSuEpGHI2K+VESeLmKc3xBR0E1EnhOR6WLjLdwTzLsOS1gTRGRCMO8EEfkm2I/viEiNArbjyjhPFC4ZVY1odnovmLcaOF5V2wLnAU/l8bx+wJOq2gb7ok4TkRbB+kcF8zOBCwvY/mnAbBGpArwCnKeqh2KVDK4SkdpYhdqDVbUVcH/kk1V1FDAd++XfRlW3RSweBZwd8fg84K0ixnkSEFme5PbgivxWQFcRaaWqT2G1fI5R1WNEpC5wB/CPYF9OB24sYDuujEvKEh6uzNsWfFlGqgg8E7TJZ2IltHP7BrhdRBpg4zD8IiLHYRVUpwXlTaqS/zgVb4rINmApNqZFc2BJRP2sV4FrsJLV6cBwEfkI+DDWN6aqa0RkcVBn55dgG18Hr1uYOKtj5SoiRyjrKSJ9sf/rfYCWWPmOSB2D+V8H26mE7Tfn8uWJwqWKAcDvWPXTctgX9U5U9b8i8h1wCjBWRPpgdX1eVdVbY9jGhao6PfuBiNTJa6WgtlB7rMhcL+BarHx1rN4CegLzgfdUVcW+tWOOExvF7UHgWeBsEWkMDASOUNX1IvIKUCWP5wowXlXPL0S8rozzpieXKnYHVgWDzfTGfk3vRESaAIuD5pYxWBPM58A5IrJXsE5tEdk/xm3OBxqJSNPgcW9gUtCmv7uqfoydKM6r59EmoGY+r/sucCZwPpY0KGycqroDa0LqGDRb7QZsATaKVUftnk8s3wJHZb8nEakmInkdnTn3N08ULlUMBS4RkW+xZqcteaxzHjBHRGZiYw28FvQ0ugMYJyKzgPFYs0yBVDUdq675TlB1NAsbOa8m8GHwepOwo53cXgGezz6Znet112NjWe+vqlODeYWOMzj38RgwUFV/BGYAc4GXseasbC8An4jIBFVdg/XIGhFs51tsXzmXL68e65xzLio/onDOOReVJwrnnHNReaJwzjkXlScK55xzUXmicM45F5UnCuecc1F5onDOORfV/wOeWMXGo99IIQAAAABJRU5ErkJggg==\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# ROC graph\n", "y_test = df.INCIDENT_2018\n", "preds = df.PREDICTION\n", "\n", "fpr, tpr, threshold = roc_curve(y_test, preds)\n", "roc_auc = auc(fpr, tpr)\n", "\n", "plt.title('Receiver Operating Characteristic')\n", "plt.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)\n", "plt.legend(loc = 'lower right')\n", "plt.plot([0, 1], [0, 1],'r--')\n", "plt.xlim([0, 1])\n", "plt.ylim([0, 1])\n", "plt.ylabel('True Positive Rate')\n", "plt.xlabel('False Positive Rate')\n", "\n", "plt.savefig('AUC ROC curve.pdf', bbox_inches = 'tight', pad_inches = 2.5)" ] }, { "cell_type": "code", "execution_count": 25, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
importance
CRIME_INCIDENT0.202975
SQFT0.185457
FRONTFOOT0.095755
NUMUNITS0.072335
TOTALVALUE0.066862
BUILDVALUE0.063689
LANDVALUE0.062413
LANDUSE0.056735
RENT_REG_UNITS0.037084
RENT_REG_YEAR0.026956
ANY_CRIME0.021229
TOTAL_VIOLATIONS0.020655
Condo or Suite0.015525
Property Maintenance0.015104
Standard Tax Parcel0.014677
ACT__MULTIFAMILY0.014435
ACT__SINGLE OR TWO RESIDENTIAL0.009917
ANY_VIOLATIONS0.007761
RENT_REG0.007503
ACT__COMMERCIAL0.002931
\n", "
" ], "text/plain": [ " importance\n", "CRIME_INCIDENT 0.202975\n", "SQFT 0.185457\n", "FRONTFOOT 0.095755\n", "NUMUNITS 0.072335\n", "TOTALVALUE 0.066862\n", "BUILDVALUE 0.063689\n", "LANDVALUE 0.062413\n", "LANDUSE 0.056735\n", "RENT_REG_UNITS 0.037084\n", "RENT_REG_YEAR 0.026956\n", "ANY_CRIME 0.021229\n", "TOTAL_VIOLATIONS 0.020655\n", "Condo or Suite 0.015525\n", "Property Maintenance 0.015104\n", "Standard Tax Parcel 0.014677\n", "ACT__MULTIFAMILY 0.014435\n", "ACT__SINGLE OR TWO RESIDENTIAL 0.009917\n", "ANY_VIOLATIONS 0.007761\n", "RENT_REG 0.007503\n", "ACT__COMMERCIAL 0.002931" ] }, "execution_count": 25, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# Feature importance from the RFC \n", "feature_importances = pd.DataFrame(random_forest.feature_importances_,\n", " index = X.columns,\n", " columns=['importance']).sort_values('importance', \n", " ascending=False)\n", "feature_importances" ] }, { "cell_type": "code", "execution_count": 26, "metadata": { "scrolled": false }, "outputs": [ { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "# Visualizing the feature importances through a graph\n", "importances = random_forest.feature_importances_\n", "indices = np.argsort(importances)\n", "ax = plt.barh(range(len(indices)), importances[indices], color='b', align='center')\n", "plt.yticks(range(len(indices)), [X.columns[i] for i in indices])\n", "plt.xlabel('Relative Importance')\n", "plt.savefig('Results - Feature importance.pdf', bbox_inches = 'tight', pad_inches = 0.5)" ] }, { "cell_type": "code", "execution_count": 27, "metadata": {}, "outputs": [], "source": [ "# Export the results to CSV\n", "df['PREDICTION'].to_csv('Results - risk predictions.csv', header = True)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## Have any questions? Email me at nsdiaz@uc.cl" ] } ], "metadata": { "celltoolbar": "Raw Cell Format", "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": "3.6.5" } }, "nbformat": 4, "nbformat_minor": 2 }