Kernel Author:
Bhishan Poudel, Ph.D Astrophysics .

Modelling Customer Churn using LogisticRegression

Go to Top

References

• Github: interpret

NOTES:

• interpret only works with built-in models: LogisticRegression,DecisionTree, and Boosting.

%%javascript
IPython.OutputArea.auto_scroll_threshold = 9999;

Load the libraries

Go to Top

import time

time_start_notebook = time.time()

Colab

%%capture
import sys
ENV_COLAB = 'google.colab' in sys.modules

if ENV_COLAB:
    # usual imports
    !pip install watermark
    !pip install scikit-plot
    !pip install alibi

    print('Environment: Google Colab')

import numpy as np
import pandas as pd
import seaborn as sns
import os,sys,time
import matplotlib.pyplot as plt
import joblib
from tqdm import tqdm, trange
import plotly_express as px

# modelling
import sklearn
import sklearn.metrics as skmetrics
from sklearn.model_selection import StratifiedKFold

# boosting
import xgboost as xgb
import lightgbm as lgb


# settings
sns.set()
SEED = 100
pd.set_option('max_columns',100)
pd.set_option('max_colwidth',200)
pd.set_option('plotting.backend','matplotlib') # matplotlib, bokeh, altair, plotly

%matplotlib inline
%load_ext watermark
%watermark -iv

json           2.0.9
sklearn        0.23.2
lightgbm       3.1.1
xgboost        1.2.0
numpy          1.19.4
seaborn        0.11.0
autopep8       1.5.4
pandas         1.1.4
plotly_express 0.4.1
joblib         0.17.0

Useful Scripts

Go to Top

def show_methods(obj, ncols=4,contains=None):
    lst = [i for i in dir(obj) if i[0]!='_' ]
    if contains is not None:
        lst = [i for i in lst if contains in i]
    df = pd.DataFrame(np.array_split(lst,ncols)).T.fillna('')
    return df

def model_eval_bin(model_name,ytest,ypreds,yprobs2d,show_plots=True):
    import sklearn.metrics as skmetrics
    import scikitplot.metrics as skpmetrics
    import os

    acc       = skmetrics.accuracy_score(ytest,ypreds)
    precision = skmetrics.precision_score(ytest,ypreds)
    recall    = skmetrics.recall_score(ytest,ypreds)
    f1        = skmetrics.f1_score(ytest,ypreds)
    auc       = skmetrics.roc_auc_score(ytest,ypreds)

    print(skmetrics.classification_report(ytest,ypreds))
    print(skmetrics.confusion_matrix(ytest,ypreds))

    df_res = pd.DataFrame({'Accuracy':[acc],
                          'Precision': [precision],
                          'Recall': [recall],
                          'F1-score': [f1],
                          'AUC': [auc]},index=[model_name])

    display(df_res.style.format("{:.4f}"))
    if not os.path.isdir('../outputs'):
        os.makedirs('../outputs')
    o = '.' if ENV_COLAB else '../outputs/'
    df_res.to_csv(o+f'model_{model_name}.csv',index=True)

    if show_plots:
        skpmetrics.plot_precision_recall(ytest,yprobs2d) # more focus on minority
        skpmetrics.plot_roc(ytest,yprobs2d) # equal focus on both groups
        skpmetrics.plot_confusion_matrix(ytest,ypreds)

def get_profit(y_true, y_pred):
    tn, fp, fn, tp = skmetrics.confusion_matrix(y_true,y_pred).ravel()
    profit = 400*tp - 200*fn - 100*fp
    return profit

scoring = skmetrics.make_scorer(get_profit, greater_is_better=True)

def set_seed(s=100):
    np.random.seed(s)
    tf.random.set_seed(s)

Load the Data

Go to Top

path_data_train = '../data/raw/train.csv'
path_data_test = '../data/raw/test.csv'

if ENV_COLAB:
    path_data_train = 'https://raw.githubusercontent.com/bhishanpdl/Datasets/master/Projects/Telco_Customer_Churn/raw/train.csv'
    path_data_test = 'https://raw.githubusercontent.com/bhishanpdl/Datasets/master/Projects/Telco_Customer_Churn/raw/test.csv'

df_train = pd.read_csv(path_data_train)
df_test = pd.read_csv(path_data_test)

print(df_train.shape)
print(df_test.shape)
df_train.head(2).append(df_train.tail(2))

(5634, 21)
(1409, 21)

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	OnlineBackup	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	1621-YNCJH	Female	0	Yes	No	36	Yes	Yes	Fiber optic	Yes	Yes	Yes	Yes	No	Yes	Two year	Yes	Credit card (automatic)	106.05	3834.4	No
1	7143-BQIBA	Male	0	No	No	10	Yes	No	DSL	Yes	No	No	Yes	Yes	No	Month-to-month	No	Bank transfer (automatic)	62.25	612.95	No
5632	0862-PRCBS	Female	0	Yes	Yes	68	Yes	Yes	Fiber optic	No	Yes	No	Yes	Yes	Yes	Two year	Yes	Credit card (automatic)	103.75	7039.45	No
5633	4656-CAURT	Male	0	No	No	69	Yes	Yes	No	No internet service	No internet service	No internet service	No internet service	No internet service	No internet service	Two year	No	Bank transfer (automatic)	23.95	1713.1	No

target_name = 'Churn'

Data Processing

Go to Top

Data Processing

sys.path.append('../src')
import util as bp_util

from sklearn.model_selection import train_test_split

df_train = bp_util.clean_data(df_train)
df_test = bp_util.clean_data(df_test)

df_train.head(2)

	customerID	tenure	MonthlyCharges	TotalCharges	Churn	Contract_TotalCharges_mean	Contract_TotalCharges_mean_diff	PaymentMethod_MonthlyCharges_mean	PaymentMethod_MonthlyCharges_mean_diff	MultipleLines_Ordinal	SeniorCitizen_Not_SenCit	SeniorCitizen_SeniorCitizen	Partner_No_Partner	Partner_Partner	Dependents_Dependents	Dependents_No_Dependents	PaperlessBilling_No_PaperlessBill	PaperlessBilling_PaperlessBill	PhoneService_No_PhoneService	PhoneService_PhoneService	OnlineSecurity_No internet service	OnlineSecurity_No_OnlineSecurity	OnlineSecurity_OnlineSecurity	OnlineBackup_No internet service	OnlineBackup_No_OnlineBackup	OnlineBackup_OnlineBackup	DeviceProtection_DeviceProtection	DeviceProtection_No internet service	DeviceProtection_No_DeviceProtection	TechSupport_No internet service	TechSupport_No_TechSupport	TechSupport_TechSupport	StreamingTV_No internet service	StreamingTV_No_StreamingTV	StreamingTV_StreamingTV	StreamingMovies_No internet service	StreamingMovies_No_StreamingMov	StreamingMovies_StreamingMov	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	Partner_Dependents_No_Partner_Dependents	Partner_Dependents_No_Partner_No_Dependents	Partner_Dependents_Partner_Dependents	Partner_Dependents_Partner_No_Dependents	SeniorCitizen_Dependents_Not_SenCit_Dependents	SeniorCitizen_Dependents_Not_SenCit_No_Dependents	SeniorCitizen_Dependents_SeniorCitizen_Dependents	SeniorCitizen_Dependents_SeniorCitizen_No_Dependents	SeniorCitizen_Partner_Not_SenCit_No_Partner	SeniorCitizen_Partner_Not_SenCit_Partner	SeniorCitizen_Partner_SeniorCitizen_No_Partner	SeniorCitizen_Partner_SeniorCitizen_Partner	SeniorCitizen_Contract_Not_SenCit_Month-to-month	SeniorCitizen_Contract_Not_SenCit_One year	SeniorCitizen_Contract_Not_SenCit_Two year	SeniorCitizen_Contract_SeniorCitizen_Month-to-month	SeniorCitizen_Contract_SeniorCitizen_One year	SeniorCitizen_Contract_SeniorCitizen_Two year	SeniorCitizen_TechSupport_Not_SenCit_No internet service	SeniorCitizen_TechSupport_Not_SenCit_No_TechSupport	SeniorCitizen_TechSupport_Not_SenCit_TechSupport	SeniorCitizen_TechSupport_SeniorCitizen_No internet service	SeniorCitizen_TechSupport_SeniorCitizen_No_TechSupport	SeniorCitizen_TechSupport_SeniorCitizen_TechSupport	SeniorCitizen_PaymentMethod_Not_SenCit_Bank transfer (automatic)	SeniorCitizen_PaymentMethod_Not_SenCit_Credit card (automatic)	SeniorCitizen_PaymentMethod_Not_SenCit_Electronic check	SeniorCitizen_PaymentMethod_Not_SenCit_Mailed check	SeniorCitizen_PaymentMethod_SeniorCitizen_Bank transfer (automatic)	SeniorCitizen_PaymentMethod_SeniorCitizen_Credit card (automatic)	SeniorCitizen_PaymentMethod_SeniorCitizen_Electronic check	SeniorCitizen_PaymentMethod_SeniorCitizen_Mailed check
0	1621-YNCJH	36	106.05	3834.40	No	3683.643192	150.756808	65.801934	40.248066	2	1	0	0	1	0	1	0	1	0	1	0	0	1	0	0	1	1	0	0	0	0	1	0	1	0	0	0	1	0	1	0	0	0	0	0	1	0	1	0	0	0	1	0	0	0	0	1	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	0	0
1	7143-BQIBA	10	62.25	612.95	No	1370.923131	-757.973131	67.564819	-5.314819	1	1	0	1	0	0	1	1	0	0	1	0	0	1	0	1	0	0	0	1	0	0	1	0	0	1	0	1	0	1	0	0	0	0	1	0	0	0	1	0	0	1	0	0	0	1	0	0	0	0	0	0	0	1	0	0	0	1	0	0	0	0	0	0	0

index_name = 'customerID'
target_name = 'Churn'
ser_train_ids = df_train.pop(index_name)
ser_test_ids = df_test.pop(index_name)


m = {'Yes':1, 'No':0}
ser_ytrain = df_train.pop(target_name).map(m)
ser_ytest = df_test.pop(target_name).map(m)

df_Xtrain_full = df_train
df_Xtest = df_test

ser_ytrain_full = ser_ytrain
ser_ytest = ser_ytest

Xtrain_full = df_Xtrain_full.to_numpy()
Xtest  = df_Xtest.to_numpy()

ytrain_full = np.array(ser_ytrain_full)
ytest = np.array(ser_ytest)

df_Xtrain,df_Xvalid,ser_ytrain,ser_yvalid = train_test_split(
    df_Xtrain_full,ser_ytrain_full,
    random_state=SEED,
    train_size=0.8
)


Xtrain = df_Xtrain.to_numpy()
Xvalid = df_Xvalid.to_numpy()

ytrain = np.array(ser_ytrain).flatten()
yvalid = np.array(ser_yvalid).flatten()

Modelling : Logistic Regression

from sklearn.linear_model import LogisticRegression

from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import PowerTransformer

df_Xtrain.head(2)

	tenure	MonthlyCharges	TotalCharges	Contract_TotalCharges_mean	Contract_TotalCharges_mean_diff	PaymentMethod_MonthlyCharges_mean	PaymentMethod_MonthlyCharges_mean_diff	MultipleLines_Ordinal	SeniorCitizen_Not_SenCit	SeniorCitizen_SeniorCitizen	Partner_No_Partner	Partner_Partner	Dependents_Dependents	Dependents_No_Dependents	PaperlessBilling_No_PaperlessBill	PaperlessBilling_PaperlessBill	PhoneService_No_PhoneService	PhoneService_PhoneService	OnlineSecurity_No internet service	OnlineSecurity_No_OnlineSecurity	OnlineSecurity_OnlineSecurity	OnlineBackup_No internet service	OnlineBackup_No_OnlineBackup	OnlineBackup_OnlineBackup	DeviceProtection_DeviceProtection	DeviceProtection_No internet service	DeviceProtection_No_DeviceProtection	TechSupport_No internet service	TechSupport_No_TechSupport	TechSupport_TechSupport	StreamingTV_No internet service	StreamingTV_No_StreamingTV	StreamingTV_StreamingTV	StreamingMovies_No internet service	StreamingMovies_No_StreamingMov	StreamingMovies_StreamingMov	PaymentMethod_Bank transfer (automatic)	PaymentMethod_Credit card (automatic)	PaymentMethod_Electronic check	PaymentMethod_Mailed check	Partner_Dependents_No_Partner_Dependents	Partner_Dependents_No_Partner_No_Dependents	Partner_Dependents_Partner_Dependents	Partner_Dependents_Partner_No_Dependents	SeniorCitizen_Dependents_Not_SenCit_Dependents	SeniorCitizen_Dependents_Not_SenCit_No_Dependents	SeniorCitizen_Dependents_SeniorCitizen_Dependents	SeniorCitizen_Dependents_SeniorCitizen_No_Dependents	SeniorCitizen_Partner_Not_SenCit_No_Partner	SeniorCitizen_Partner_Not_SenCit_Partner	SeniorCitizen_Partner_SeniorCitizen_No_Partner	SeniorCitizen_Partner_SeniorCitizen_Partner	SeniorCitizen_Contract_Not_SenCit_Month-to-month	SeniorCitizen_Contract_Not_SenCit_One year	SeniorCitizen_Contract_Not_SenCit_Two year	SeniorCitizen_Contract_SeniorCitizen_Month-to-month	SeniorCitizen_Contract_SeniorCitizen_One year	SeniorCitizen_Contract_SeniorCitizen_Two year	SeniorCitizen_TechSupport_Not_SenCit_No internet service	SeniorCitizen_TechSupport_Not_SenCit_No_TechSupport	SeniorCitizen_TechSupport_Not_SenCit_TechSupport	SeniorCitizen_TechSupport_SeniorCitizen_No internet service	SeniorCitizen_TechSupport_SeniorCitizen_No_TechSupport	SeniorCitizen_TechSupport_SeniorCitizen_TechSupport	SeniorCitizen_PaymentMethod_Not_SenCit_Bank transfer (automatic)	SeniorCitizen_PaymentMethod_Not_SenCit_Credit card (automatic)	SeniorCitizen_PaymentMethod_Not_SenCit_Electronic check	SeniorCitizen_PaymentMethod_Not_SenCit_Mailed check	SeniorCitizen_PaymentMethod_SeniorCitizen_Bank transfer (automatic)	SeniorCitizen_PaymentMethod_SeniorCitizen_Credit card (automatic)	SeniorCitizen_PaymentMethod_SeniorCitizen_Electronic check	SeniorCitizen_PaymentMethod_SeniorCitizen_Mailed check
973	47	110.85	5275.8	3018.965636	2256.834364	65.801934	45.048066	2	1	0	0	1	0	1	0	1	0	1	0	0	1	0	1	0	1	0	0	0	0	1	0	0	1	0	0	1	0	1	0	0	0	0	0	1	0	1	0	0	0	1	0	0	0	1	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	0	0
54	47	58.60	2723.4	1370.923131	1352.476869	67.564819	-8.964819	1	1	0	0	1	0	1	0	1	0	1	0	1	0	0	0	1	0	0	1	0	1	0	0	0	1	0	1	0	1	0	0	0	0	0	0	1	0	1	0	0	0	1	0	0	1	0	0	0	0	0	0	1	0	0	0	0	1	0	0	0	0	0	0	0

# Define the columns we wish to transform
cols_scale = ['tenure', 'MonthlyCharges', 'TotalCharges',
              'Contract_TotalCharges_mean',
              'Contract_TotalCharges_mean_diff',
              'PaymentMethod_MonthlyCharges_mean',
             'PaymentMethod_MonthlyCharges_mean_diff']

# Scale the relevant columns
transformer = ColumnTransformer([('yeo_johnson', PowerTransformer(), cols_scale)], 
                                remainder='passthrough')
transformer.fit(df_Xtrain)

df_Xtrain_scaled = pd.DataFrame(transformer.transform(df_Xtrain))
df_Xtest_scaled = pd.DataFrame(transformer.transform(df_Xtest))

#features
df_Xtrain_scaled.columns = df_Xtrain.columns
df_Xtest_scaled.columns = df_Xtest.columns

params = {'C': 0.42679058013626753, 'max_iter': 1000,
               'penalty': 'l2', 'solver': 'lbfgs'}

model = LogisticRegression(**params)
model.fit(df_Xtrain, ser_ytrain)
ypreds   = model.predict(df_Xtest)

profit = get_profit(ytest,ypreds)
print(f'profit without scaled = ${profit:,d}')

profit without scaled = $36,100

model = LogisticRegression(**params)
model.fit(df_Xtrain_scaled, ser_ytrain)
ypreds   = model.predict(df_Xtest_scaled)

profit = get_profit(ytest,ypreds)
print(f'profit with scaled = ${profit:,d}')

profit with scaled = $37,200

Model interpretation: interpret

import interpret
from interpret.data import ClassHistogram

show_methods(interpret)

	0	1	2	3
0	NullHandler	init_show_server	set_visualize_provider	status_show_server
1	api	logging	show	utils
2	data	preserve	show_link	version
3	get_show_addr	provider	shutdown_show_server	visual
4	get_visualize_provider	set_show_addr

show_methods(interpret.data)

	0	1	2	3
0	ClassHistogram	Marginal	response

Class Histogram

hist = ClassHistogram().explain_data(df_Xtrain, ytrain, name = 'Train Data')
interpret.show(hist)

Interpret LogisticRegression

from interpret import glassbox
show_methods(glassbox)

	0	1	2	3
0	ClassificationTree	ExplainableBoostingRegressor	RegressionTree	linear
1	DecisionListClassifier	LinearRegression	decisiontree	skoperules
2	ExplainableBoostingClassifier	LogisticRegression	ebm

model = glassbox.LogisticRegression(**params)
model.fit(df_Xtrain_scaled,ytrain)

<interpret.glassbox.linear.LogisticRegression at 0x7fd9ffdee110>

show_methods(model)

	0	1	2	3
0	X_maxs_	categorical_uniq_	feature_types	predict
1	X_mins_	explain_global	fit	predict_proba
2	available_explanations	explain_local	global_selector	score
3	bin_counts_	explainer_type	kwargs	sk_model_
4	bin_edges_	feature_names	linear_class

Interpret: explain global

model_global = model.explain_global(name='logistic regression')
interpret.show(model_global)

Interpret: explain local

model_local = model.explain_local(
    df_Xtest.iloc[:5], ytrain[:5],
    name='logistic regression')
interpret.show(model_local)

Model performance: ROC

from interpret import perf as in_perf

show_methods(in_perf,5)

	0	1	2	3	4
0	PR	ROC	RegressionPerf	curve	regression

e = in_perf.ROC(model.predict_proba)
e = e.explain_perf(df_Xtest,ytest,
                    name='Logistic Regression')

interpret.show(e)

Compare logistic regression with Boosting

glassbox.ExplainableBoostingClassifier(
feature_names            = None,
feature_types            = None,
max_bins                 = 255,
max_interaction_bins     = 32,
binning                  = 'quantile',
mains                    = 'all',
interactions             = 0,
outer_bags               = 16,
inner_bags               = 0,
learning_rate            = 0.01,
validation_size          = 0.15,
early_stopping_rounds    = 50,
early_stopping_tolerance = 0.0001,
max_rounds               = 5000,
max_leaves               = 3,
min_samples_leaf         = 2,
n_jobs                   = -2,
random_state             = 42,
)

# glassbox.ExplainableBoostingClassifier?

model_gbm = glassbox.ExplainableBoostingClassifier()
model_gbm.fit(df_Xtrain_scaled,ytrain);

lr_perf = (in_perf.ROC(model.predict_proba)
           .explain_perf(df_Xtest, ytest,
                         name='Logistic Regression')
          )

gbm_perf = (in_perf.ROC(model_gbm.predict_proba)
             .explain_perf(df_Xtest, ytest,
                           name='Boosting')
           )

lr_global = model.explain_global(name='Logistic Regression')
gbm_global = model_gbm.explain_global(name='Boosting')

interpret.show([hist,
                lr_global, lr_perf,
                gbm_global, gbm_perf],
               share_tables=True)
            

Open in new window

Interpret Shap

from interpret import blackbox
show_methods(blackbox)

	0	1	2	3
0	LimeTabular	PartialDependence	lime	sensitivity
1	MorrisSensitivity	ShapKernel	partialdependence	shap

cols_all = df_Xtrain.columns.tolist()

from interpret.blackbox import ShapKernel


background_val = np.median(Xtrain, axis=0).reshape(1, -1)

shap = ShapKernel(predict_fn=model.predict_proba,
                  data=background_val,
                  feature_names=cols_all)


shap_local = shap.explain_local(Xtest[:5], ytest[:5],
                                name='SHAP')
interpret.show(shap_local)

Global Explanations: How the model behaves overall

from interpret.blackbox import MorrisSensitivity

sensitivity = MorrisSensitivity(
    predict_fn=model.predict_proba, data=Xtrain)

sensitivity_global = sensitivity.explain_global(
    name="Global Sensitivity")

interpret.show(sensitivity_global)

from interpret.blackbox import PartialDependence

pdp = PartialDependence(
    predict_fn=model.predict_proba, data=Xtrain)
pdp_global = pdp.explain_global(name='Partial Dependence')

interpret.show(pdp_global)

interpret.show([shap_local, sensitivity_global, pdp_global])

Open in new window