Telco Customer Churn Cohort Study

Github Repository

• Feature Exploration
• Churn Analaysis
  • Tenure Cohort Study
• Predictive Modeling
  • Single Decision Tree
  • Random Forest
  • Adaptive Boosting
  • Gradient Boosting
  • Support Vector Classifier

import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
from sklearn import svm
from sklearn.ensemble import (
    RandomForestClassifier,
    GradientBoostingClassifier,
    AdaBoostClassifier
)
from sklearn.metrics import (
    classification_report,
    confusion_matrix,
    ConfusionMatrixDisplay,
)
from sklearn.model_selection import (
    train_test_split,
    GridSearchCV
)
from sklearn.tree import DecisionTreeClassifier
from sklearn.preprocessing import StandardScaler

!wget https://github.com/mpolinowski/datasets/raw/master/telco_customer_churn.csv -P dataset

telco_df = pd.read_csv('dataset/telco_customer_churn.csv')
telco_df.head(5).transpose()

	0	1	2	3	4
gender	Female	Male	Male	Male	Female
SeniorCitizen	0	0	0	0	0
Partner	Yes	No	No	No	No
Dependents	No	No	No	No	No
tenure	1	34	2	45	2
PhoneService	No	Yes	Yes	No	Yes
MultipleLines	No phone service	No	No	No phone service	No
InternetService	DSL	DSL	DSL	DSL	Fiber optic
OnlineSecurity	No	Yes	Yes	Yes	No
OnlineBackup	Yes	No	Yes	No	No
DeviceProtection	No	Yes	No	Yes	No
TechSupport	No	No	No	Yes	No
StreamingTV	No	No	No	No	No
StreamingMovies	No	No	No	No	No
Contract	Month-to-month	One year	Month-to-month	One year	Month-to-month
PaperlessBilling	Yes	No	Yes	No	Yes
PaymentMethod	Electronic check	Mailed check	Mailed check	Bank transfer (automatic)	Electronic check
MonthlyCharges	29.85	56.95	53.85	42.3	70.7
TotalCharges	29.85	1889.5	108.15	1840.75	151.65
Churn	No	No	Yes	No	Yes

Feature Exploration

# check for missing data or wrong dtype
telco_df.info()

<class 'pandas.core.frame.DataFrame'>

RangeIndex: 7043 entries, 0 to 7042 Data columns (total 20 columns):

#	Column	Non-Null Count	Dtype
0	gender	7043 non-null	object
1	SeniorCitizen	7043 non-null	int64
2	Partner	7043 non-null	object
3	Dependents	7043 non-null	object
4	tenure	7043 non-null	int64
5	PhoneService	7043 non-null	object
6	MultipleLines	7043 non-null	object
7	InternetService	7043 non-null	object
8	OnlineSecurity	7043 non-null	object
9	OnlineBackup	7043 non-null	object
10	DeviceProtection	7043 non-null	object
11	TechSupport	7043 non-null	object
12	StreamingTV	7043 non-null	object
13	StreamingMovies	7043 non-null	object
14	Contract	7043 non-null	object
15	PaperlessBilling	7043 non-null	object
16	PaymentMethod	7043 non-null	object
17	MonthlyCharges	7043 non-null	float64
18	TotalCharges	7043 non-null	object
19	Churn	7043 non-null	object
dtypes: float64(1), int64(2), object(17)
memory usage: 1.1+ MB

# TotalCharges is type string because it contains SPACE's
# how many rows have a SPACE instad of a float value?
(telco_df['TotalCharges'] == ' ').sum()
# 11

# drop those 11 rows
telco_df = telco_df[telco_df['TotalCharges'] != ' ']

# now turn TotalCharges column into dtype=float
telco_df['TotalCharges'] = telco_df['TotalCharges'].apply(lambda string: float(string))
# Name: TotalCharges, Length: 7032, dtype: float64

# check for missing values
telco_df.isnull().sum()

gender	0
SeniorCitizen	0
Partner	0
Dependents	0
tenure	0
PhoneService	0
MultipleLines	0
InternetService	0
OnlineSecurity	0
OnlineBackup	0
DeviceProtection	0
TechSupport	0
StreamingTV	0
StreamingMovies	0
Contract	0
PaperlessBilling	0
PaymentMethod	0
MonthlyCharges	0
TotalCharges	0
Churn	0
dtype: int64

# check label distribution
telco_df['Churn'].value_counts()

# No     5174
# Yes    1869
# Name: Churn, dtype: int64

plt.figure(figsize=(10,4))
plt.title('Churn Count Distribution')
sns.countplot(data=telco_df, x='Churn', palette='seismic')
plt.savefig('assets/telco-customer-churn_01.webp', bbox_inches='tight')

plt.figure(figsize=(10, 5))
plt.title('Total Charges by Churn Count')

sns.boxenplot(
    data=telco_df,
    x='Churn',
    y='TotalCharges',
    hue='gender',
    palette='seismic_r'
)
plt.savefig('assets/telco-customer-churn_02.webp', bbox_inches='tight')

plt.figure(figsize=(10, 5))
plt.title('Total Charges by Contract Type')

sns.violinplot(
    data=telco_df,
    x='Contract',
    y='TotalCharges',
    hue='Churn',
    palette='seismic_r',
    orient='v',
    inner='quartile',
    bw=0.3,
    split=True
)
plt.legend(loc='lower right', title='Churn')
plt.savefig('assets/telco-customer-churn_03.webp', bbox_inches='tight')

# make values numeric to show correlation
telco_df_numeric = pd.get_dummies(telco_df[
    [
        'gender',
        'SeniorCitizen',
        'Partner',
        'Dependents',
        'PhoneService',
        'MultipleLines',
        'InternetService',
        'OnlineSecurity',
        'OnlineBackup',
        'DeviceProtection',
        'TechSupport',
        'StreamingTV',
        'StreamingMovies',
        'Contract',
        'PaperlessBilling',
        'PaymentMethod',
        'Churn'
    ]
], drop_first=True)

telco_df_numeric.head(5).transpose()

	0	1	2	3	4
SeniorCitizen	0	0	0	0	0
gender_Male	0	1	1	1	0
Partner_Yes	1	0	0	0	0
Dependents_Yes	0	0	0	0	0
PhoneService_Yes	0	1	1	0	1
MultipleLines_No phone service	1	0	0	1	0
MultipleLines_Yes	0	0	0	0	0
InternetService_Fiber optic	0	0	0	0	1
InternetService_No	0	0	0	0	0
OnlineSecurity_No internet service	0	0	0	0	0
OnlineSecurity_Yes	0	1	1	1	0
OnlineBackup_No internet service	0	0	0	0	0
OnlineBackup_Yes	1	0	1	0	0
DeviceProtection_No internet service	0	0	0	0	0
DeviceProtection_Yes	0	1	0	1	0
TechSupport_No internet service	0	0	0	0	0
TechSupport_Yes	0	0	0	1	0
StreamingTV_No internet service	0	0	0	0	0
StreamingTV_Yes	0	0	0	0	0
StreamingMovies_No internet service	0	0	0	0	0
StreamingMovies_Yes	0	0	0	0	0
Contract_One year	0	1	0	1	0
Contract_Two year	0	0	0	0	0
PaperlessBilling_Yes	1	0	1	0	1
PaymentMethod_Credit card (automatic)	0	0	0	0	0
PaymentMethod_Electronic check	1	0	0	0	1
PaymentMethod_Mailed check	0	1	1	0	0
Churn_Yes	0	0	1	0	1

plt.figure(figsize=(25,25))
plt.title('Dataset Correlation Matrix')
sns.heatmap(telco_df_numeric.corr(), annot=True, cmap='coolwarm')
plt.savefig('assets/telco-customer-churn_04.webp', bbox_inches='tight')

# correlation to label class
telco_df_numeric.corr(numeric_only=True)['Churn_Yes']

SeniorCitizen	0.150541
gender_Male	-0.008545
Partner_Yes	-0.149982
Dependents_Yes	-0.163128
PhoneService_Yes	0.011691
MultipleLines_No phone service	-
MultipleLines_Yes	0.040033
InternetService_Fiber optic	0.307463
InternetService_No	-0.227578
OnlineSecurity_No internet service	-0.227578
OnlineSecurity_Yes	-0.171270
OnlineBackup_No internet service	-0.227578
OnlineBackup_Yes	-0.082307
DeviceProtection_No internet service	-0.227578
DeviceProtection_Yes	-0.066193
TechSupport_No internet service	-0.227578
TechSupport_Yes	-0.164716
StreamingTV_No internet service	-0.227578
StreamingTV_Yes	0.063254
StreamingMovies_No internet service	-0.227578
StreamingMovies_Yes	0.060860
Contract_One year	-0.178225
Contract_Two year	-0.301552
PaperlessBilling_Yes	0.191454
PaymentMethod_Credit card (automatic	-0.134687
PaymentMethod_Electronic check	0.301455
PaymentMethod_Mailed check	-0.090773
Churn_Yes	1.000000
Name: Churn_Yes, dtype: float64

# create correlation dataframe
telco_df_numeric_corr = telco_df_numeric.corr(numeric_only=True)['Churn_Yes'][:-2].sort_values()

plt.figure(figsize=(10, 10))
plt.title('Festure Correlation to Customer Churn')

sns.barplot(
    y=telco_df_numeric_corr.index,
    x=telco_df_numeric_corr.values,
    estimator='mean',
    errorbar=None,
    palette='coolwarm',
    orient='h'
)

plt.savefig('assets/telco-customer-churn_05.webp', bbox_inches='tight')

Churn Analaysis

# segmenting customers based on their tenure
telco_df['Contract'].value_counts()

# Month-to-month    3875
# Two year          1685
# One year          1472
# Name: Contract, dtype: int64

plt.figure(figsize=(10, 5))
plt.title('Churn Distribution by Contract Type')

sns.histplot(
    data=telco_df,
    x='Contract',
    hue='Churn',
    element='step',
    palette='seismic',
    kde=True
)

plt.savefig('assets/telco-customer-churn_06.webp', bbox_inches='tight')

plt.figure(figsize=(10, 5))
plt.title('Churn Distribution by Tenure in Months')

sns.histplot(
    data=telco_df,
    x='tenure',
    hue='Churn',
    element='step',
    palette='seismic',
    kde=True
)

plt.savefig('assets/telco-customer-churn_07.webp', bbox_inches='tight')

grid = sns.catplot(
    data=telco_df,
    x='tenure',
    kind='count',
    hue='Churn',
    palette='seismic',
    col='Contract',
    row='Churn'
)

grid.set(
    xticks=np.arange(0,72,6),
    xlabel='Tenure [Months]',
    ylabel='Customer Count'
)

grid.fig.subplots_adjust(top=0.9)
grid.fig.suptitle('Tenure [months] by Contract Type and Churn')

plt.savefig('assets/telco-customer-churn_08.webp', bbox_inches='tight')

plt.figure(figsize=(10, 6))

plot = sns.scatterplot(
    x='MonthlyCharges',
    y='TotalCharges',
    data=telco_df,
    hue='Churn',
    palette='seismic',
    alpha=0.5
)

plot.set_title('Churn by Monthly and Total Charges')

plt.savefig('assets/telco-customer-churn_09.webp', bbox_inches='tight')

Tenure Cohort Study

# group customers by tenure and calculate average monthly charges
telco_df_tenure = telco_df.groupby('tenure').mean(numeric_only=True)['MonthlyCharges']

tenure	MonthlyCharges
1	50.485808
2	57.206303
3	58.015000
4	57.432670
5	61.003759
...
68	73.321000
69	70.823158
70	76.378992
71	73.735588
72	80.695856
Name: MonthlyCharges, Length: 72, dtype: float64

plt.figure(figsize=(10, 5))
plt.title('Average Monthly Charges by Tenure')
plt.xlabel('Tenure [Months]')
plt.ylabel('Mean Monthly Charges [$]')

sns.scatterplot(
    x=telco_df_tenure.index,
    y=telco_df_tenure.values,
    color='#3b70f7'
)

plt.savefig('assets/telco-customer-churn_10.webp', bbox_inches='tight')

# adding missing columns
telco_df_numeric['Tenure'] = telco_df['tenure']
telco_df_numeric['Monthly Charges'] = telco_df['MonthlyCharges']
telco_df_numeric['Total Charges'] = telco_df['TotalCharges']
telco_df_numeric['Contract'] = telco_df['Contract']
telco_df_numeric[['Tenure', 'Churn_Yes']]

	Tenure	Churn_Yes
0	1	0
1	34	0
2	2	1
3	45	0
4	2	1
...
7038	24	0
7039	72	0
7040	11	0
7041	4	1
7042	66	0
7032 rows × 2 columns

# group customers by tenure and calculate average churn rate
telco_df_tenure_Churn = telco_df_numeric.groupby('Tenure').mean(numeric_only=True)['Churn_Yes']

Tenure	Average Churn Rate
1	0.619902
2	0.516807
3	0.470000
4	0.471591
5	0.481203
...
68	0.090000
69	0.084211
70	0.092437
71	0.035294
72	0.016575
Name: Churn_Yes, Length: 72, dtype: float64

fig, ax = plt.subplots(
    figsize=(10,5)
)
ax.bar(
    x=telco_df_tenure_Churn.index,
    height=telco_df_tenure_Churn.values.round(4)*100,
    color='#3b70f7'
)
ax.set_title('Churn Rate by Tenure in Months')
ax.set(
    xlabel='Tenure [Months]',
    ylabel='Churn Rate [%]',
    xticks=np.arange(0,78,6)
)

fig.savefig('assets/telco-customer-churn_11.webp', bbox_inches='tight')

# group customers in chorts 0-12,12-24,24-48 and above 48 months tenure
def tenure_cohort(tenure):
    if tenure < 13:
        return '0-12 Months'
    elif tenure < 25:
        return '12-24 Months'
    elif tenure < 49:
        return '24-48 Months'
    else:
        return 'Over 48 Months'
        

telco_df_numeric['Tenure Cohort'] = telco_df_numeric['Tenure'].apply(tenure_cohort)
telco_df_numeric[['Tenure','Tenure Cohort']]

	Tenure	Tenure Cohort
0	1	0-12 Months
1	34	24-48 Months
2	2	0-12 Months
3	45	24-48 Months
4	2	0-12 Months
...
7038	24	12-24 Months
7039	72	Over 48 Months
7040	11	0-12 Months
7041	4	0-12 Months
7042	66	Over 48 Months
7032 rows × 2 columns

plt.figure(figsize=(10, 6))

plot = sns.scatterplot(
    x='Monthly Charges',
    y='Total Charges',
    data=telco_df_numeric,
    hue='Tenure Cohort',
    palette='seismic_r',
    alpha=0.5
)

plot.set_title('Monthly and Total Charges by Tenure Cohort')

plt.savefig('assets/telco-customer-churn_12.webp', bbox_inches='tight')

plt.figure(figsize=(12, 5))
plt.title('Monthly Charges for Tenure Cohorts')

sns.boxenplot(
    data=telco_df_numeric,
    x='Tenure Cohort',
    y='Monthly Charges',
    hue='Churn_Yes',
    palette='seismic_r'
)
plt.savefig('assets/telco-customer-churn_13.webp', bbox_inches='tight')

plt.figure(figsize=(10, 5))
plt.title('Total Charges for Tenure Cohorts')

sns.boxenplot(
    data=telco_df_numeric,
    x='Tenure Cohort',
    y='Total Charges',
    hue='Churn_Yes',
    palette='seismic_r'
)
plt.savefig('assets/telco-customer-churn_14.webp', bbox_inches='tight')

plt.figure(figsize=(10, 5))
plt.title('Churn by Tenure Cohorts')

sns.countplot(
    data=telco_df_numeric,
    x='Tenure Cohort',
    hue='Churn_Yes',
    palette='seismic_r'
)
plt.savefig('assets/telco-customer-churn_15.webp', bbox_inches='tight')

grid = sns.catplot(
    data=telco_df_numeric,
    x='Tenure Cohort',
    kind='count',
    hue='Churn_Yes',
    palette='seismic',
    col='Contract'
)

grid.set(
    # xticks=np.arange(0,72,6),
    xlabel='Tenure Cohort',
    ylabel='Customer Count'
)

grid.fig.subplots_adjust(top=0.9)
grid.fig.suptitle('Contract Type by Tenure Cohort and Churn')

plt.savefig('assets/telco-customer-churn_16.webp', bbox_inches='tight')

Predictive Modeling

Single Decision Tree

Baseline single tree classifier with default hyper parameters.

# remove target feature and non-numeric columns from training set
X_churn = telco_df_numeric.drop(['Churn_Yes','Tenure Cohort', 'Contract'], axis=1)
y_churn = telco_df_numeric['Churn_Yes']

# train/test split
X_churn_train, X_churn_test, y_churn_train, y_churn_test = train_test_split(
    X_churn,
    y_churn,
    test_size=0.2
)

# fit the model
tree_classifier = DecisionTreeClassifier()
tree_classifier.fit(X_churn_train, y_churn_train)

# run predictions
y_tree_preds = tree_classifier.predict(X_churn_test)

# evaluate predictions
conf_mtx_tree = confusion_matrix(y_churn_test, y_tree_preds)

conf_mtx_tree_plot = ConfusionMatrixDisplay(
    confusion_matrix=conf_mtx_tree,
    display_labels=[False,True]
)

conf_mtx_tree_plot.plot(cmap='winter')

plt.savefig('assets/telco-customer-churn_17.webp', bbox_inches='tight')

report_tree = classification_report(
    y_churn_test, y_tree_preds
)
print(report_tree)

	precision	recall	f1-score	support
0	0.80	0.81	0.81	1012
1	0.51	0.50	0.50	395
accuracy			0.72	1407
macro avg	0.66	0.65	0.65	1407
weighted avg	0.72	0.72	0.72	1407

# feature importance for classification
tree_index = ['importance']
tree_data_columns = pd.Series(X_churn.columns)
tree_importance_array = tree_classifier.feature_importances_
tree_importance_df = pd.DataFrame(tree_importance_array, tree_data_columns, tree_index)
tree_importance_df

	importance
SeniorCitizen	0.019747
gender_Male	0.022160
Partner_Yes	0.018070
Dependents_Yes	0.016844
PhoneService_Yes	0.002513
MultipleLines_No phone service	0.001701
MultipleLines_Yes	0.015637
InternetService_Fiber optic	0.106278
InternetService_No	0.000000
OnlineSecurity_No internet service	0.000000
OnlineSecurity_Yes	0.015796
OnlineBackup_No internet service	0.000000
OnlineBackup_Yes	0.015036
DeviceProtection_No internet service	0.000000
DeviceProtection_Yes	0.013911
TechSupport_No internet service	0.004314
TechSupport_Yes	0.023056
StreamingTV_No internet service	0.006463
StreamingTV_Yes	0.012094
StreamingMovies_No internet service	0.000000
StreamingMovies_Yes	0.010126
Contract_One year	0.008873
Contract_Two year	0.007519
PaperlessBilling_Yes	0.023079
PaymentMethod_Credit card (automatic)	0.010816
PaymentMethod_Electronic check	0.022332
PaymentMethod_Mailed check	0.009718
Tenure	0.192801
MonthlyCharges	0.085796
TotalCharges	0.121363
Monthly Charges	0.114721
Total Charges	0.099236

tree_importance_df.sort_values(
    by='importance',
    ascending=True
).plot(
    kind='barh',
    title='Feature Importance for Churn Classification',
    figsize=(10,10)
)

plt.savefig('assets/telco-customer-churn_18.webp', bbox_inches='tight')

Random Forest

Random forest with grid search for optimal hyper parameter.

rdnfor_classifier = RandomForestClassifier()

param_grid = {
    'n_estimators': [5, 25, 50, 75,100, 125],
    'min_samples_split': [1,2,3],
    'min_samples_leaf': [1,2,3],
    'criterion': ['gini', 'entropy', 'log_loss'],
    'max_features' : ['sqrt', 'log2']
}

grid_search_rdnfor = GridSearchCV(
    estimator = rdnfor_classifier,
    param_grid = param_grid
)

grid_search_rdnfor.fit(X_churn_train, y_churn_train)

print('Best Parameter: ', grid_search_rdnfor.best_params_)
# Best Parameter:  {'criterion': 'gini', 'max_features': 'log2', 'min_samples_leaf': 3, 'min_samples_split': 2, 'n_estimators': 125}

# run predictions
y_rdnfor_preds = grid_search_rdnfor.predict(X_churn_test)

# evaluate predictions
conf_mtx_rdnfor = confusion_matrix(y_churn_test, y_rdnfor_preds)

conf_mtx_rdnfor_plot = ConfusionMatrixDisplay(
    confusion_matrix=conf_mtx_rdnfor,
    display_labels=[False,True]
)

conf_mtx_rdnfor_plot.plot(cmap='winter')

plt.savefig('assets/telco-customer-churn_19.webp', bbox_inches='tight')

report_rdnfor = classification_report(
    y_churn_test, y_rdnfor_preds
)
print(report_rdnfor)

	precision	recall	f1-score	support
0	0.84	0.91	0.87	1012
1	0.70	0.55	0.61	395
accuracy			0.81	1407
macro avg	0.77	0.73	0.74	1407
weighted avg	0.80	0.81	0.80	1407

Adaptive Boosting

Baseline AdaBoost classifier with default hyper parameters.

abc_classifier = AdaBoostClassifier(
    estimator=None,
    n_estimators=50,
    learning_rate=1.0
)

abc_classifier.fit(X_churn_train, y_churn_train)

abc_classifier_preds = abc_classifier.predict(X_churn_test)

# run predictions
y_abc_preds = abc_classifier.predict(X_churn_test)

# evaluate predictions
conf_mtx_abc = confusion_matrix(y_churn_test, y_abc_preds)

conf_mtx_abc_plot = ConfusionMatrixDisplay(
    confusion_matrix=conf_mtx_abc,
    display_labels=[False,True]
)

conf_mtx_abc_plot.plot(cmap='winter')

plt.savefig('assets/telco-customer-churn_21.webp', bbox_inches='tight')

report_abc = classification_report(
    y_churn_test, y_abc_preds
)
print(report_abc)

	precision	recall	f1-score	support
0	0.85	0.91	0.88	1012
1	0.72	0.58	0.64	395
accuracy			0.82	1407
macro avg	0.78	0.75	0.76	1407
weighted avg	0.81	0.82	0.81	1407

# feature importance for classification
abc_index = ['importance']
abc_data_columns = pd.Series(X_churn.columns)
abc_importance_array = abc_classifier.feature_importances_
abc_importance_df = pd.DataFrame(abc_importance_array, abc_data_columns, abc_index)
abc_importance_df

abc_importance_df.sort_values(
    by='importance',
    ascending=True
).plot(
    kind='barh',
    title='Feature Importance for Churn Classification',
    figsize=(10,10)
)

plt.savefig('assets/telco-customer-churn_22.webp', bbox_inches='tight')

Gradient Boosting

GradientBoost Classifier with grid search for optimal hyper parameter.

gb_classifier = GradientBoostingClassifier()

param_grid = {
    'n_estimators': [50, 100, 150],
    'learning_rate': [0.05,0.1,0.2],
    'max_depth': [2,3,4,5]
}

grid_search_gb = GridSearchCV(gb_classifier, param_grid)
grid_search_gb.fit(X_churn_train, y_churn_train)
grid_search_gb.best_params_
# {'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 100}

# run predictions
y_gb_preds = grid_search_gb.predict(X_churn_test)

# evaluate predictions
conf_mtx_gb = confusion_matrix(y_churn_test, y_gb_preds)

conf_mtx_gb_plot = ConfusionMatrixDisplay(
    confusion_matrix=conf_mtx_gb,
    display_labels=[False,True]
)

conf_mtx_gb_plot.plot(cmap='winter')

plt.savefig('assets/telco-customer-churn_23.webp', bbox_inches='tight')

report_gb = classification_report(
    y_churn_test, y_gb_preds
)
print(report_gb)

	precision	recall	f1-score	support
0	0.84	0.92	0.88	1012
1	0.73	0.56	0.63	395
accuracy			0.82	1407
macro avg	0.79	0.74	0.75	1407
weighted avg	0.81	0.82	0.81	1407

Support Vector Classifier

# normalize feature values
sc_churn = StandardScaler()
X_churn_train_scaled=sc_churn.fit_transform(X_churn_train)
X_churn_test_scaled=sc_churn.transform(X_churn_test)

# SVM classifier fitting
svc_classifier = svm.SVC()
svc_classifier.fit(X_churn_train_scaled, y_churn_train)

# Predictions
y_svc_pred = svc_classifier.predict(X_churn_test_scaled)

# evaluate predictions
conf_mtx_svc = confusion_matrix(y_churn_test, y_svc_pred)

conf_mtx_svc_plot = ConfusionMatrixDisplay(
    confusion_matrix=conf_mtx_svc,
    display_labels=[False,True]
)

conf_mtx_svc_plot.plot(cmap='winter')

plt.savefig('assets/telco-customer-churn_24.webp', bbox_inches='tight')

report_svc = classification_report(
    y_churn_test, y_svc_pred
)
print(report_svc)

	precision	recall	f1-score	support
0	0.83	0.92	0.87	1012
1	0.72	0.50	0.59	395
accuracy			0.81	1407
macro avg	0.77	0.71	0.73	1407
weighted avg	0.80	0.81	0.79	1407