import pandas as pd # data manipulation and analysis
import numpy as np # numerical
import matplotlib.pyplot as plt # visualisations
import seaborn as sns # visualisation (advanced)
from sklearn.model_selection import train_test_split, GridSearchCV # model selector/tuner
from sklearn.preprocessing import StandardScaler, LabelEncoder # data preprosessing
from sklearn.linear_model import LogisticRegression # logistic regression
from sklearn.svm import SVC # Support Vector Classifier
from sklearn.metrics import (accuracy_score, # model performance tools
                            precision_score, 
                            recall_score, 
                            f1_score, 
                            roc_auc_score, 
                            roc_curve)
from sklearn.metrics import confusion_matrix # confusion metrics

df = pd.read_excel('Telco_Customer_Churn.xlsx')

df.head(10)

df.tail(10)

df.describe()

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7043 entries, 0 to 7042
Data columns (total 33 columns):
 #   Column             Non-Null Count  Dtype  
---  ------             --------------  -----  
 0   CustomerID         7043 non-null   object 
 1   Count              7043 non-null   int64  
 2   Country            7043 non-null   object 
 3   State              7043 non-null   object 
 4   City               7043 non-null   object 
 5   Zip Code           7043 non-null   int64  
 6   Lat Long           7043 non-null   object 
 7   Latitude           7043 non-null   float64
 8   Longitude          7043 non-null   float64
 9   Gender             7043 non-null   object 
 10  Senior Citizen     7043 non-null   object 
 11  Partner            7043 non-null   object 
 12  Dependents         7043 non-null   object 
 13  Tenure Months      7043 non-null   int64  
 14  Phone Service      7043 non-null   object 
 15  Multiple Lines     7043 non-null   object 
 16  Internet Service   7043 non-null   object 
 17  Online Security    7043 non-null   object 
 18  Online Backup      7043 non-null   object 
 19  Device Protection  7043 non-null   object 
 20  Tech Support       7043 non-null   object 
 21  Streaming TV       7043 non-null   object 
 22  Streaming Movies   7043 non-null   object 
 23  Contract           7043 non-null   object 
 24  Paperless Billing  7043 non-null   object 
 25  Payment Method     7043 non-null   object 
 26  Monthly Charges    7043 non-null   float64
 27  Total Charges      7043 non-null   object 
 28  Churn Label        7043 non-null   object 
 29  Churn Value        7043 non-null   int64  
 30  Churn Score        7043 non-null   int64  
 31  CLTV               7043 non-null   int64  
 32  Churn Reason       1869 non-null   object 
dtypes: float64(3), int64(6), object(24)
memory usage: 1.8+ MB

df.nunique()

CustomerID           7043
Count                   1
Country                 1
State                   1
City                 1129
Zip Code             1652
Lat Long             1652
Latitude             1652
Longitude            1651
Gender                  2
Senior Citizen          2
Partner                 2
Dependents              2
Tenure Months          73
Phone Service           2
Multiple Lines          3
Internet Service        3
Online Security         3
Online Backup           3
Device Protection       3
Tech Support            3
Streaming TV            3
Streaming Movies        3
Contract                3
Paperless Billing       2
Payment Method          4
Monthly Charges      1585
Total Charges        6531
Churn Label             2
Churn Value             2
Churn Score            85
CLTV                 3438
Churn Reason           20
dtype: int64

df = df.drop(['Churn Value','Lat Long','CustomerID','City','Count', 'Country', 'State', 'Zip Code', 'Latitude', 'Longitude'], axis=1)

num_cols=['Tenure Months','Monthly Charges','Churn Score','CLTV','Total Charges']

cat_cols=['Gender','Senior Citizen','Partner','Dependents','Phone Service','Multiple Lines','Internet Service','Online Security','Online Backup','Device Protection','Tech Support','Streaming TV','Streaming Movies','Contract','Paperless Billing','Payment Method','Churn Label','Churn Reason']

# coerce means that numeric values in the column are changed to NaN
df['Total Charges'] = pd.to_numeric(df['Total Charges'], errors='coerce') 
# NaN values change to the column mean value
df['Total Charges'] = df['Total Charges'].fillna(df['Total Charges'].mean())

df_model = df.copy()

# pd.get_dummies changes categorical into indicators from cat_cols
# drop_first=True drops the first category of each cat_cols variable to stop variable trap
df_encoded = pd.get_dummies(df, columns=cat_cols, drop_first=True)

# Drop churn label from independent variables
X = df_encoded.drop('Churn Label_Yes', axis=1)
# Make churn label the dependent variable
y = df_encoded['Churn Label_Yes']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, # 20% testing
                 random_state=42, # reproducibility with fixed seed for random splitting
                 stratify=y) # proportioned split for target and training

scaler = StandardScaler() # mean = 0 and standard deviation = 1
X_train[num_cols] = scaler.fit_transform(X_train[num_cols]) # scale training data
X_test[num_cols] = scaler.transform(X_test[num_cols]) # scale numerical columns without fitting for consistency

df[num_cols].describe().T

# printing proportions for each category
for i in cat_cols:
    print(df[i].value_counts (normalize=True))
    print('*'*40)

Gender
Male      0.504756
Female    0.495244
Name: proportion, dtype: float64
****************************************
Senior Citizen
No     0.837853
Yes    0.162147
Name: proportion, dtype: float64
****************************************
Partner
No     0.516967
Yes    0.483033
Name: proportion, dtype: float64
****************************************
Dependents
No     0.76899
Yes    0.23101
Name: proportion, dtype: float64
****************************************
Phone Service
Yes    0.903166
No     0.096834
Name: proportion, dtype: float64
****************************************
Multiple Lines
No                  0.481329
Yes                 0.421837
No phone service    0.096834
Name: proportion, dtype: float64
****************************************
Internet Service
Fiber optic    0.439585
DSL            0.343746
No             0.216669
Name: proportion, dtype: float64
****************************************
Online Security
No                     0.496663
Yes                    0.286668
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Online Backup
No                     0.438450
Yes                    0.344881
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Device Protection
No                     0.439443
Yes                    0.343888
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Tech Support
No                     0.493114
Yes                    0.290217
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Streaming TV
No                     0.398978
Yes                    0.384353
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Streaming Movies
No                     0.395428
Yes                    0.387903
No internet service    0.216669
Name: proportion, dtype: float64
****************************************
Contract
Month-to-month    0.550192
Two year          0.240664
One year          0.209144
Name: proportion, dtype: float64
****************************************
Paperless Billing
Yes    0.592219
No     0.407781
Name: proportion, dtype: float64
****************************************
Payment Method
Electronic check             0.335794
Mailed check                 0.228880
Bank transfer (automatic)    0.219225
Credit card (automatic)      0.216101
Name: proportion, dtype: float64
****************************************
Churn Label
No     0.73463
Yes    0.26537
Name: proportion, dtype: float64
****************************************
Churn Reason
Attitude of support person                   0.102729
Competitor offered higher download speeds    0.101124
Competitor offered more data                 0.086677
Don't know                                   0.082397
Competitor made better offer                 0.074906
Attitude of service provider                 0.072231
Competitor had better devices                0.069556
Network reliability                          0.055110
Product dissatisfaction                      0.054575
Price too high                               0.052434
Service dissatisfaction                      0.047619
Lack of self-service on Website              0.047084
Extra data charges                           0.030498
Moved                                        0.028357
Limited range of services                    0.023542
Long distance charges                        0.023542
Lack of affordable download/upload speed     0.023542
Poor expertise of phone support              0.010701
Poor expertise of online support             0.010166
Deceased                                     0.003210
Name: proportion, dtype: float64
****************************************

df[num_cols].describe().T # summary statistics for numerical columns, transpose rows to columns
df[num_cols].hist(figsize=(14, 14)) # histogram creation for numerical columns
plt.show() # show histograms

for i in cat_cols:
    # exclude churn label
    if i!='Churn Label':
        # create normalised contingency table of columns and churn label, shown as percentage of bar, 
        (pd.crosstab(df[i],df['Churn Label'],normalize='index')*100).plot(kind='bar',figsize=(8,4),stacked=True)
        # set y axis label to %
        plt.ylabel('%')
        # display tables
        plt.show()

# Checking the relationship between different numerical variables

# Plotting the correlation between numerical variables

plt.figure(figsize=(15,8)) # set plot size
sns.heatmap(df[num_cols].corr(), annot=True, fmt='0.2f', cmap='YlOrRd') # correlation for numerical columns, displaying 2d.p and colour themed
plt.title('Corrolation Matrix of Numerical Variables') # title heat map
plt.show() # show heatmap

# start logistic regression with fixed random state (more reliable)
log_reg = LogisticRegression(random_state=42)
# fit feature matrix and target vector
log_reg.fit(X_train, y_train)

LogisticRegression(random_state=42)

LogisticRegression(random_state=42)

# make predictions using trained logistic regression, on feature matrix and target values for test set
y_pred_lr = log_reg.predict(X_test)

# proportion of correctly predicted instances out of all instances
accuracy = accuracy_score(y_test, y_pred_lr) 
# proportion of true positive instances out of all instances predicted
precision = precision_score(y_test, y_pred_lr)
# proportion of true positive instances out of all actual positive instances
recall = recall_score(y_test, y_pred_lr)
# f1 score of all predictions
f1 = f1_score(y_test, y_pred_lr)
# ROC AUC score of predictions
roc_auc = roc_auc_score(y_test, y_pred_lr)

# print all results
print(f'Accuracy: {accuracy:.4f}')
print(f'Precision: {precision:.4f}')
print(f'Recall: {recall:.4f}')
print(f'F1 Score: {f1:.4f}')
print(f'ROC AUC: {roc_auc:.4f}')

Accuracy: 0.9879
Precision: 0.9837
Recall: 0.9706
F1 Score: 0.9771
ROC AUC: 0.9824

# apply confusion matrix onto the logistical regression model
conf_matrix = confusion_matrix(y_test, y_pred_lr)

# set plot size
plt.figure(figsize=(10, 6))

#create  heatmap of confusion matrix, with annotations, blue colours, and labeled axis
sns.heatmap(conf_matrix, annot=True, fmt='d', cmap='Blues', xticklabels=['Not Churn', 'Churn'], yticklabels=['Not Churn', 'Churn'])
plt.xlabel('Predicted')
plt.ylabel('Actual')
plt.title('Confusion Matrix')

# show confusion matrix
plt.show()

# enables probability estimates which is needed for ROC AUC analysis
svm = SVC(probability=True)
# fits SVC model to training data, on feature matrix and target vector training set
svm.fit(X_train, y_train)

SVC(probability=True)

SVC(probability=True)

# Make predictions on the test set
y_pred_svm = svm.predict(X_test)

# Get predicted probabilities for the positive class (churn)
y_prob_svm = svm.predict_proba(X_test)[:, 1]

# Accuracy
accuracy_svm = accuracy_score(y_test, y_pred_svm)
print(f"Accuracy: {accuracy_svm:.4f}")

# Precision
precision_svm = precision_score(y_test, y_pred_svm)
print(f"Precision: {precision_svm:.4f}")

# Recall
recall_svm = recall_score(y_test, y_pred_svm)
print(f"Recall: {recall_svm:.4f}")

# F1 Score
f1_svm = f1_score(y_test, y_pred_svm)
print(f"F1 Score: {f1_svm:.4f}")

# ROC AUC (using predicted probabilities for the positive class)
roc_auc_svm = roc_auc_score(y_test, y_prob_svm)
print(f"ROC AUC: {roc_auc_svm:.4f}")

Accuracy: 0.9851
Precision: 0.9862
Recall: 0.9572
F1 Score: 0.9715
ROC AUC: 0.9989

# Generate confusion matrix
cm_svm = confusion_matrix(y_test, y_pred_svm)

# Visualize the confusion matrix using a heatmap
plt.figure(figsize=(6,6))
sns.heatmap(cm_svm, annot=True, fmt='d', cmap='Blues', xticklabels=['Non-Churn', 'Churn'], yticklabels=['Non-Churn', 'Churn'])
plt.ylabel('Actual')
plt.xlabel('Predicted')
plt.title('Confusion Matrix for SVM')
plt.show()

def evaluate_model(y_test, y_pred, model_name):
    accuracy = accuracy_score(y_test, y_pred)
    precision = precision_score(y_test, y_pred)
    recall = recall_score(y_test, y_pred)
    f1 = f1_score(y_test, y_pred)
    auc_roc = roc_auc_score(y_test, y_pred)
    
    print(f"{model_name} Model Performance:")
    print(f"Accuracy: {accuracy}")
    print(f"Precision: {precision}")
    print(f"Recall: {recall}")
    print(f"F1 Score: {f1}")
    print(f"AUC-ROC: {auc_roc}")
    print("\n")

evaluate_model(y_test, y_pred_lr, "Logistic Regression")
evaluate_model(y_test, y_pred_svm, "Support Vector Machine")

Logistic Regression Model Performance:
Accuracy: 0.9879347054648687
Precision: 0.983739837398374
Recall: 0.9705882352941176
F1 Score: 0.9771197846567967
AUC-ROC: 0.9823955669224211


Support Vector Machine Model Performance:
Accuracy: 0.9850958126330731
Precision: 0.9862258953168044
Recall: 0.9572192513368984
F1 Score: 0.9715061058344641
AUC-ROC: 0.9761941667312511

# Initialize the Support Vector Machine classifier with probability estimation
svm = SVC(probability=True, random_state=42)

# Fit the SVM to the training data
svm.fit(X_train, y_train)

# Make predictions on the test set
y_pred_svm = svm.predict(X_test)

param_grid = {
    'C': [0.1, 1, 10, 100],  # Regularization parameter
    'kernel': ['linear', 'rbf', 'poly'],  # Kernel type
    'gamma': ['scale', 'auto'],  # Kernel coefficient
    'class_weight': [None, 'balanced']  # Handle class imbalance
}

grid_search = GridSearchCV(svm, param_grid, cv=5, verbose=1, n_jobs=-1, scoring='accuracy')

grid_search.fit(X_train, y_train)

Fitting 5 folds for each of 48 candidates, totalling 240 fits

GridSearchCV(cv=5, estimator=SVC(probability=True, random_state=42), n_jobs=-1,
             param_grid={'C': [0.1, 1, 10, 100],
                         'class_weight': [None, 'balanced'],
                         'gamma': ['scale', 'auto'],
                         'kernel': ['linear', 'rbf', 'poly']},
             scoring='accuracy', verbose=1)

GridSearchCV(cv=5, estimator=SVC(probability=True, random_state=42), n_jobs=-1,
             param_grid={'C': [0.1, 1, 10, 100],
                         'class_weight': [None, 'balanced'],
                         'gamma': ['scale', 'auto'],
                         'kernel': ['linear', 'rbf', 'poly']},
             scoring='accuracy', verbose=1)

SVC(probability=True, random_state=42)

SVC(probability=True, random_state=42)

# Best hyperparameters found from GridSearchCV
print(f"Best hyperparameters found: {grid_search.best_params_}")

# Best model from grid search
best_svm = grid_search.best_estimator_
print(best_svm)

Best hyperparameters found: {'C': 10, 'class_weight': None, 'gamma': 'scale', 'kernel': 'linear'}
SVC(C=10, kernel='linear', probability=True, random_state=42)

# Make predictions using the best model
y_pred_best_svm = best_svm.predict(X_test)

# Evaluate performance metrics
print("SVM Model Performance (Best Model from Grid Search):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_best_svm):.4f}")
print(f"Precision: {precision_score(y_test, y_pred_best_svm):.4f}")
print(f"Recall: {recall_score(y_test, y_pred_best_svm):.4f}")
print(f"F1 Score: {f1_score(y_test, y_pred_best_svm):.4f}")
print(f"ROC AUC: {roc_auc_score(y_test, best_svm.predict_proba(X_test)[:, 1]):.4f}")

SVM Model Performance (Best Model from Grid Search):
Accuracy: 0.9922
Precision: 1.0000
Recall: 0.9706
F1 Score: 0.9851
ROC AUC: 0.9987

# Generate the confusion matrix
cm = confusion_matrix(y_test, y_pred_best_svm)

# Plot the confusion matrix
plt.figure(figsize=(6, 5))
sns.heatmap(cm, annot=True, fmt="d", cmap="Blues", xticklabels=["No Churn", "Churn"], yticklabels=["No Churn", "Churn"])
plt.title("Confusion Matrix")
plt.xlabel("Predicted")
plt.ylabel("Actual")
plt.show()

# Define a reduced parameter grid for Logistic Regression
# There are three dictionaries
# C gives regularisaiton strength because of smaller values
param_grid_lr = [
    # solver, Liblinear is a optimisation algorithm suitable for small datasets and both l1 and l2
    {'C': [0.1, 1, 10], 'solver': ['liblinear'], 'penalty': ['l1', 'l2']},
    # solver, ibfgs is an optimisation algorithm suitable for l2 penalty only
    {'C': [0.1, 1, 10], 'solver': ['lbfgs'], 'penalty': ['l2']},
    # solver saga is an optimisation algorithm suitable for l1 penalty only
    {'C': [0.1, 1, 10], 'solver': ['saga'], 'penalty': ['l1', 'l2']}
]

grid_search_lr = GridSearchCV(LogisticRegression(random_state=42), param_grid_lr, cv=5, verbose=1, n_jobs=-1, scoring='accuracy')
grid_search_lr.fit(X_train, y_train)

Fitting 5 folds for each of 15 candidates, totalling 75 fits

GridSearchCV(cv=5, estimator=LogisticRegression(random_state=42), n_jobs=-1,
             param_grid=[{'C': [0.1, 1, 10], 'penalty': ['l1', 'l2'],
                          'solver': ['liblinear']},
                         {'C': [0.1, 1, 10], 'penalty': ['l2'],
                          'solver': ['lbfgs']},
                         {'C': [0.1, 1, 10], 'penalty': ['l1', 'l2'],
                          'solver': ['saga']}],
             scoring='accuracy', verbose=1)

GridSearchCV(cv=5, estimator=LogisticRegression(random_state=42), n_jobs=-1,
             param_grid=[{'C': [0.1, 1, 10], 'penalty': ['l1', 'l2'],
                          'solver': ['liblinear']},
                         {'C': [0.1, 1, 10], 'penalty': ['l2'],
                          'solver': ['lbfgs']},
                         {'C': [0.1, 1, 10], 'penalty': ['l1', 'l2'],
                          'solver': ['saga']}],
             scoring='accuracy', verbose=1)

LogisticRegression(random_state=42)

LogisticRegression(random_state=42)

# Best Logistic Regression model from grid search
best_log_reg = grid_search_lr.best_estimator_
y_pred_best_log_reg = best_log_reg.predict(X_test)

# Evaluate performance metrics for the best Logistic Regression model
accuracy_best_log_reg = accuracy_score(y_test, y_pred_best_log_reg)
precision_best_log_reg = precision_score(y_test, y_pred_best_log_reg)
recall_best_log_reg = recall_score(y_test, y_pred_best_log_reg)
f1_best_log_reg = f1_score(y_test, y_pred_best_log_reg)
roc_auc_best_log_reg = roc_auc_score(y_test, y_pred_best_log_reg)

print(f'Logistic Regression Model Performance (Best Model from Reduced Grid Search):')
print(f'Accuracy: {accuracy_best_log_reg:.4f}')
print(f'Precision: {precision_best_log_reg:.4f}')
print(f'Recall: {recall_best_log_reg:.4f}')
print(f'F1 Score: {f1_best_log_reg:.4f}')
print(f'ROC AUC: {roc_auc_best_log_reg:.4f}')

Logistic Regression Model Performance (Best Model from Reduced Grid Search):
Accuracy: 0.9943
Precision: 0.9946
Recall: 0.9840
F1 Score: 0.9892
ROC AUC: 0.9910

# Ensure that the best Logistic Regression model is defined
best_log_reg = grid_search_lr.best_estimator_

# Make predictions using the best Logistic Regression model
y_pred_best_log_reg = best_log_reg.predict(X_test)

print(f'Logistic Regression Model Performance (Best Model from Grid Search):')
print(f'Accuracy: {accuracy_best_log_reg:.4f}')
print(f'Precision: {precision_best_log_reg:.4f}')
print(f'Recall: {recall_best_log_reg:.4f}')
print(f'F1 Score: {f1_best_log_reg:.4f}')
print(f'ROC AUC: {roc_auc_best_log_reg:.4f}')

Logistic Regression Model Performance (Best Model from Grid Search):
Accuracy: 0.9943
Precision: 0.9946
Recall: 0.9840
F1 Score: 0.9892
ROC AUC: 0.9910

# Cross-validation for Logistic Regression
#cv_scores_lr_after = cross_val_score(best_log_reg, X, y, cv=5, scoring='accuracy')
#print(f"Logistic Regression Cross-Validation Accuracy (After Tuning): {cv_scores_lr_after.mean():.4f}")

# Step 1: Evaluate Logistic Regression before and after hyperparameter tuning
log_reg_baseline = LogisticRegression(random_state=42).fit(X_train, y_train)
y_pred_log_reg_baseline = log_reg_baseline.predict(X_test)

# Step 2: Evaluate SVM before and after hyperparameter tuning
y_pred_svm_baseline = svm.predict(X_test)

# Step 3: Hyperparameter tuning for both models (as done earlier)

# Compare performance before and after tuning
print(f"Logistic Regression (Before Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_log_reg_baseline)}")
print(f"Precision: {precision_score(y_test, y_pred_log_reg_baseline)}")
print(f"Recall: {recall_score(y_test, y_pred_log_reg_baseline)}")
print(f"F1 Score: {f1_score(y_test, y_pred_log_reg_baseline)}")

print(f"\nSVM (Before Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_svm_baseline)}")
print(f"Precision: {precision_score(y_test, y_pred_svm_baseline)}")
print(f"Recall: {recall_score(y_test, y_pred_svm_baseline)}")
print(f"F1 Score: {f1_score(y_test, y_pred_svm_baseline)}")

Logistic Regression (Before Tuning):
Accuracy: 0.9879347054648687
Precision: 0.983739837398374
Recall: 0.9705882352941176
F1 Score: 0.9771197846567967

SVM (Before Tuning):
Accuracy: 0.9850958126330731
Precision: 0.9862258953168044
Recall: 0.9572192513368984
F1 Score: 0.9715061058344641

print("\nLogistic Regression (After Tuning):")
print(f"Accuracy: {accuracy_best_log_reg:.4f}")
print(f"Precision: {precision_best_log_reg:.4f}")
print(f"Recall: {recall_best_log_reg:.4f}")
print(f"F1 Score: {f1_best_log_reg:.4f}")

Logistic Regression (After Tuning):
Accuracy: 0.9943
Precision: 0.9946
Recall: 0.9840
F1 Score: 0.9892

print("\nSVM (After Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_best_svm):.4f}")
print(f"Precision: {precision_score(y_test, y_pred_best_svm):.4f}")
print(f"Recall: {recall_score(y_test, y_pred_best_svm):.4f}")
print(f"F1 Score: {f1_score(y_test, y_pred_best_svm):.4f}")

SVM (After Tuning):
Accuracy: 0.9922
Precision: 1.0000
Recall: 0.9706
F1 Score: 0.9851

# Performance metrics before and after tuning
metrics = ['Accuracy', 'Precision', 'Recall', 'F1 Score', 'ROC AUC']
log_reg_before = [accuracy_score(y_test, y_pred_log_reg_baseline), precision_score(y_test, y_pred_log_reg_baseline), recall_score(y_test, y_pred_log_reg_baseline), f1_score(y_test, y_pred_log_reg_baseline), roc_auc_score(y_test, y_pred_log_reg_baseline)]
log_reg_after = [accuracy_best_log_reg, precision_best_log_reg, recall_best_log_reg, f1_best_log_reg, roc_auc_best_log_reg]
svm_before = [accuracy_score(y_test, y_pred_svm_baseline), precision_score(y_test, y_pred_svm_baseline), recall_score(y_test, y_pred_svm_baseline), f1_score(y_test, y_pred_svm_baseline), roc_auc_score(y_test, y_pred_svm_baseline)]
svm_after = [accuracy_score(y_test, y_pred_best_svm), precision_score(y_test, y_pred_best_svm), recall_score(y_test, y_pred_best_svm), f1_score(y_test, y_pred_best_svm), roc_auc_score(y_test, best_svm.predict_proba(X_test)[:, 1])]

# Plotting the metrics
fig, ax = plt.subplots(1, 2, figsize=(14, 6))

ax[0].bar(metrics, log_reg_before, color='blue', alpha=0.6, label='Before Tuning')
ax[0].bar(metrics, log_reg_after, color='green', alpha=0.6, label='After Tuning')
ax[0].set_title('Logistic Regression Performance')
ax[0].legend()

ax[1].bar(metrics, svm_before, color='blue', alpha=0.6, label='Before Tuning')
ax[1].bar(metrics, svm_after, color='green', alpha=0.6, label='After Tuning')
ax[1].set_title('SVM Performance')
ax[1].legend()

plt.show()

# Create a summary table
comparison_data = {
    'Metric': metrics,
    'LR (Before Tuning)': log_reg_before,
    'LR (After Tuning)': log_reg_after,
    'SVM (Before Tuning)': svm_before,
    'SVM (After Tuning)': svm_after
}

comparison_df = pd.DataFrame(comparison_data)
print(comparison_df)

      Metric  LR (Before Tuning)  LR (After Tuning)  SVM (Before Tuning)  \
0   Accuracy            0.987935           0.994322             0.985096   
1  Precision            0.983740           0.994595             0.986226   
2     Recall            0.970588           0.983957             0.957219   
3   F1 Score            0.977120           0.989247             0.971506   
4    ROC AUC            0.982396           0.991012             0.976194   

   SVM (After Tuning)  
0            0.992193  
1            1.000000  
2            0.970588  
3            0.985075  
4            0.998695

# Summary of Findings
print("Logistic Regression (Before Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_log_reg_baseline):.4f}")
print(f"Precision: {precision_score(y_test, y_pred_log_reg_baseline):.4f}")
print(f"Recall: {recall_score(y_test, y_pred_log_reg_baseline):.4f}")
print(f"F1 Score: {f1_score(y_test, y_pred_log_reg_baseline):.4f}")

Logistic Regression (Before Tuning):
Accuracy: 0.9879
Precision: 0.9837
Recall: 0.9706
F1 Score: 0.9771

print("\nLogistic Regression (After Tuning):")
print(f"Accuracy: {accuracy_best_log_reg:.4f}")
print(f"Precision: {precision_best_log_reg:.4f}")
print(f"Recall: {recall_best_log_reg:.4f}")
print(f"F1 Score: {f1_best_log_reg:.4f}")

Logistic Regression (After Tuning):
Accuracy: 0.9943
Precision: 0.9946
Recall: 0.9840
F1 Score: 0.9892

print("\nSVM (Before Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_svm_baseline):.4f}")
print(f"Precision: {precision_score(y_test, y_pred_svm_baseline):.4f}")
print(f"Recall: {recall_score(y_test, y_pred_svm_baseline):.4f}")
print(f"F1 Score: {f1_score(y_test, y_pred_svm_baseline):.4f}")

SVM (Before Tuning):
Accuracy: 0.9851
Precision: 0.9862
Recall: 0.9572
F1 Score: 0.9715

print("\nSVM (After Tuning):")
print(f"Accuracy: {accuracy_score(y_test, y_pred_best_svm):.4f}")
print(f"Precision: {precision_score(y_test, y_pred_best_svm):.4f}")
print(f"Recall: {recall_score(y_test, y_pred_best_svm):.4f}")
print(f"F1 Score: {f1_score(y_test, y_pred_best_svm):.4f}")

SVM (After Tuning):
Accuracy: 0.9922
Precision: 1.0000
Recall: 0.9706
F1 Score: 0.9851

# Feature Importance for Logistic Regression
feature_importance_lr = pd.Series(best_log_reg.coef_[0], index=X.columns).sort_values(ascending=False)
print("Feature Importance for Logistic Regression:")
print(feature_importance_lr)

Feature Importance for Logistic Regression:
Churn Reason_Long distance charges                        11.917486
Churn Reason_Competitor offered more data                 10.821918
Churn Reason_Competitor offered higher download speeds    10.429499
Churn Reason_Attitude of support person                   10.342916
Churn Reason_Competitor made better offer                 10.311438
Churn Reason_Don't know                                   10.231137
Churn Reason_Price too high                                9.995710
Churn Reason_Network reliability                           9.836814
Churn Reason_Moved                                         9.805171
Churn Reason_Lack of self-service on Website               9.552347
Churn Reason_Product dissatisfaction                       9.228050
Churn Reason_Service dissatisfaction                       9.027239
Churn Reason_Competitor had better devices                 8.845489
Churn Reason_Extra data charges                            8.545527
Churn Reason_Lack of affordable download/upload speed      7.706435
Churn Reason_Poor expertise of online support              7.513258
Churn Reason_Poor expertise of phone support               7.428125
Churn Reason_Limited range of services                     7.268097
Churn Reason_Deceased                                      6.422142
Churn Score                                                5.026159
Monthly Charges                                            1.100508
Partner_Yes                                                0.603408
Total Charges                                              0.583322
Paperless Billing_Yes                                      0.485778
Senior Citizen_Yes                                         0.455502
Payment Method_Electronic check                            0.261504
CLTV                                                       0.171309
Internet Service_Fiber optic                               0.054558
Contract_One year                                          0.030911
Streaming Movies_Yes                                       0.000000
Streaming Movies_No internet service                       0.000000
Streaming TV_No internet service                           0.000000
Tech Support_No internet service                           0.000000
Device Protection_Yes                                      0.000000
Device Protection_No internet service                      0.000000
Online Backup_No internet service                          0.000000
Online Security_No internet service                        0.000000
Internet Service_No                                        0.000000
Multiple Lines_Yes                                        -0.051372
Payment Method_Credit card (automatic)                    -0.117536
Payment Method_Mailed check                               -0.159349
Streaming TV_Yes                                          -0.163500
Online Security_Yes                                       -0.218493
Gender_Male                                               -0.253644
Tech Support_Yes                                          -0.399558
Contract_Two year                                         -0.691048
Online Backup_Yes                                         -0.703037
Dependents_Yes                                            -0.870152
Multiple Lines_No phone service                           -1.814781
Tenure Months                                             -1.954257
Phone Service_Yes                                         -3.591200
dtype: float64

# Permutation Importance for SVM
from sklearn.inspection import permutation_importance
perm_importance = permutation_importance(best_svm, X_test, y_test)
feature_importance_svm = pd.Series(perm_importance.importances_mean, index=X.columns).sort_values(ascending=False)
print("Feature Importance for SVM:")
print(feature_importance_svm)

Feature Importance for SVM:
Churn Score                                               0.162243
Churn Reason_Attitude of support person                   0.027963
Churn Reason_Don't know                                   0.022427
Churn Reason_Competitor offered more data                 0.021434
Churn Reason_Competitor offered higher download speeds    0.019730
Churn Reason_Competitor had better devices                0.017317
Churn Reason_Competitor made better offer                 0.016040
Churn Reason_Price too high                               0.015188
Churn Reason_Network reliability                          0.011923
Churn Reason_Extra data charges                           0.009226
Churn Reason_Lack of self-service on Website              0.008091
Churn Reason_Service dissatisfaction                      0.007807
Churn Reason_Product dissatisfaction                      0.007381
Churn Reason_Limited range of services                    0.006529
Churn Reason_Long distance charges                        0.005820
Churn Reason_Moved                                        0.004826
Churn Reason_Lack of affordable download/upload speed     0.004400
Churn Reason_Poor expertise of phone support              0.001845
Tenure Months                                             0.000852
Churn Reason_Poor expertise of online support             0.000710
Multiple Lines_Yes                                        0.000284
Device Protection_Yes                                     0.000142
Contract_One year                                         0.000142
Payment Method_Mailed check                               0.000142
Streaming Movies_Yes                                      0.000142
Streaming TV_Yes                                          0.000142
Monthly Charges                                           0.000000
Churn Reason_Deceased                                     0.000000
Total Charges                                             0.000000
CLTV                                                      0.000000
Gender_Male                                               0.000000
Tech Support_Yes                                          0.000000
Dependents_Yes                                            0.000000
Phone Service_Yes                                         0.000000
Multiple Lines_No phone service                           0.000000
Streaming TV_No internet service                          0.000000
Streaming Movies_No internet service                      0.000000
Internet Service_No                                       0.000000
Online Security_No internet service                       0.000000
Tech Support_No internet service                          0.000000
Online Backup_No internet service                         0.000000
Online Backup_Yes                                         0.000000
Device Protection_No internet service                     0.000000
Payment Method_Credit card (automatic)                    0.000000
Contract_Two year                                         0.000000
Online Security_Yes                                       0.000000
Internet Service_Fiber optic                             -0.000142
Partner_Yes                                              -0.000284
Payment Method_Electronic check                          -0.000426
Senior Citizen_Yes                                       -0.000568
Paperless Billing_Yes                                    -0.000568
dtype: float64

# Actionable Insights and Recommendations
print("\nActionable Insights:")
print("1. Customer Retention: Focus on improving services and offering loyalty programs.")
print("2. Targeted Marketing: Identify high-risk customers and target them with personalized campaigns.")
print("3. Service Improvement: Enhance services that are associated with higher churn.")

Actionable Insights:
1. Customer Retention: Focus on improving services and offering loyalty programs.
2. Targeted Marketing: Identify high-risk customers and target them with personalized campaigns.
3. Service Improvement: Enhance services that are associated with higher churn.

print("\nRecommendations:")
print("1. Model Deployment: Deploy the best-performing model for real-time churn prediction.")
print("2. Continuous Monitoring: Regularly monitor and retrain the model.")
print("3. Customer Feedback: Collect and address feedback from high-risk customers.")

Recommendations:
1. Model Deployment: Deploy the best-performing model for real-time churn prediction.
2. Continuous Monitoring: Regularly monitor and retrain the model.
3. Customer Feedback: Collect and address feedback from high-risk customers.

	CustomerID	Count	Country	State	City	Zip Code	Lat Long	Latitude	Longitude	Gender	...	Contract	Paperless Billing	Payment Method	Monthly Charges	Total Charges	Churn Label	Churn Value	Churn Score	CLTV	Churn Reason
0	3668-QPYBK	1	United States	California	Los Angeles	90003	33.964131, -118.272783	33.964131	-118.272783	Male	...	Month-to-month	Yes	Mailed check	53.85	108.15	Yes	1	86	3239	Competitor made better offer
1	9237-HQITU	1	United States	California	Los Angeles	90005	34.059281, -118.30742	34.059281	-118.307420	Female	...	Month-to-month	Yes	Electronic check	70.70	151.65	Yes	1	67	2701	Moved
2	9305-CDSKC	1	United States	California	Los Angeles	90006	34.048013, -118.293953	34.048013	-118.293953	Female	...	Month-to-month	Yes	Electronic check	99.65	820.5	Yes	1	86	5372	Moved
3	7892-POOKP	1	United States	California	Los Angeles	90010	34.062125, -118.315709	34.062125	-118.315709	Female	...	Month-to-month	Yes	Electronic check	104.80	3046.05	Yes	1	84	5003	Moved
4	0280-XJGEX	1	United States	California	Los Angeles	90015	34.039224, -118.266293	34.039224	-118.266293	Male	...	Month-to-month	Yes	Bank transfer (automatic)	103.70	5036.3	Yes	1	89	5340	Competitor had better devices
5	4190-MFLUW	1	United States	California	Los Angeles	90020	34.066367, -118.309868	34.066367	-118.309868	Female	...	Month-to-month	No	Credit card (automatic)	55.20	528.35	Yes	1	78	5925	Competitor offered higher download speeds
6	8779-QRDMV	1	United States	California	Los Angeles	90022	34.02381, -118.156582	34.023810	-118.156582	Male	...	Month-to-month	Yes	Electronic check	39.65	39.65	Yes	1	100	5433	Competitor offered more data
7	1066-JKSGK	1	United States	California	Los Angeles	90024	34.066303, -118.435479	34.066303	-118.435479	Male	...	Month-to-month	No	Mailed check	20.15	20.15	Yes	1	92	4832	Competitor made better offer
8	6467-CHFZW	1	United States	California	Los Angeles	90028	34.099869, -118.326843	34.099869	-118.326843	Male	...	Month-to-month	Yes	Electronic check	99.35	4749.15	Yes	1	77	5789	Competitor had better devices
9	8665-UTDHZ	1	United States	California	Los Angeles	90029	34.089953, -118.294824	34.089953	-118.294824	Male	...	Month-to-month	No	Electronic check	30.20	30.2	Yes	1	97	2915	Competitor had better devices

	CustomerID	Count	Country	State	City	Zip Code	Lat Long	Latitude	Longitude	Gender	...	Contract	Paperless Billing	Payment Method	Monthly Charges	Total Charges	Churn Label	Churn Score	CLTV	Churn Reason
7033	0871-OPBXW	1	United States	California	Twentynine Palms	92277	34.17211, -115.769773	34.172110	-115.769773	Female	...	Month-to-month	Yes	Mailed check	20.05	39.25	No	80	5191	NaN
7034	3605-JISKB	1	United States	California	Twentynine Palms	92278	34.457829, -116.139589	34.457829	-116.139589	Male	...	One year	No	Credit card (automatic)	60.00	3316.1	No	71	4212	NaN
7035	9767-FFLEM	1	United States	California	Westmorland	92281	33.03679, -115.60503	33.036790	-115.605030	Male	...	Month-to-month	Yes	Credit card (automatic)	69.50	2625.25	No	35	4591	NaN
7036	8456-QDAVC	1	United States	California	Winterhaven	92283	32.852947, -114.850784	32.852947	-114.850784	Male	...	Month-to-month	Yes	Bank transfer (automatic)	78.70	1495.1	No	20	2464	NaN
7037	7750-EYXWZ	1	United States	California	Yucca Valley	92284	34.159534, -116.425984	34.159534	-116.425984	Female	...	One year	No	Electronic check	60.65	743.3	No	24	3740	NaN
7038	2569-WGERO	1	United States	California	Landers	92285	34.341737, -116.539416	34.341737	-116.539416	Female	...	Two year	Yes	Bank transfer (automatic)	21.15	1419.4	No	45	5306	NaN
7039	6840-RESVB	1	United States	California	Adelanto	92301	34.667815, -117.536183	34.667815	-117.536183	Male	...	One year	Yes	Mailed check	84.80	1990.5	No	59	2140	NaN
7040	2234-XADUH	1	United States	California	Amboy	92304	34.559882, -115.637164	34.559882	-115.637164	Female	...	One year	Yes	Credit card (automatic)	103.20	7362.9	No	71	5560	NaN
7041	4801-JZAZL	1	United States	California	Angelus Oaks	92305	34.1678, -116.86433	34.167800	-116.864330	Female	...	Month-to-month	Yes	Electronic check	29.60	346.45	No	59	2793	NaN
7042	3186-AJIEK	1	United States	California	Apple Valley	92308	34.424926, -117.184503	34.424926	-117.184503	Male	...	Two year	Yes	Bank transfer (automatic)	105.65	6844.5	No	38	5097	NaN

	Count	Zip Code	Latitude	Longitude	Tenure Months	Monthly Charges	Churn Value	Churn Score	CLTV
count	7043.0	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000	7043.000000
mean	1.0	93521.964646	36.282441	-119.798880	32.371149	64.761692	0.265370	58.699418	4400.295755
std	0.0	1865.794555	2.455723	2.157889	24.559481	30.090047	0.441561	21.525131	1183.057152
min	1.0	90001.000000	32.555828	-124.301372	0.000000	18.250000	0.000000	5.000000	2003.000000
25%	1.0	92102.000000	34.030915	-121.815412	9.000000	35.500000	0.000000	40.000000	3469.000000
50%	1.0	93552.000000	36.391777	-119.730885	29.000000	70.350000	0.000000	61.000000	4527.000000
75%	1.0	95351.000000	38.224869	-118.043237	55.000000	89.850000	1.000000	75.000000	5380.500000
max	1.0	96161.000000	41.962127	-114.192901	72.000000	118.750000	1.000000	100.000000	6500.000000

	count	mean	std	min	25%	50%	75%	max
Tenure Months	7043.0	32.371149	24.559481	0.00	9.000	29.00	55.00	72.00
Monthly Charges	7043.0	64.761692	30.090047	18.25	35.500	70.35	89.85	118.75
Churn Score	7043.0	58.699418	21.525131	5.00	40.000	61.00	75.00	100.00
CLTV	7043.0	4400.295755	1183.057152	2003.00	3469.000	4527.00	5380.50	6500.00
Total Charges	7043.0	2283.300441	2265.000258	18.80	402.225	1400.55	3786.60	8684.80

1. Inserting data¶

Import libraries¶

Insert the file using the path location¶

2. Data exploration¶

Check the first 10 rows of the dataset¶

Check the last 10 rows of the dataset¶

Summary of numerical features¶

Info about dataset (check data types and missing values)¶

Unique values¶

3. Data Cleaning¶

Drop columns¶

Creating numerical columns¶

Creating categorical variables¶

Convert 'Total Charges' to numeric and handle errors¶

Create a copy of the DataFrame for model training¶

One-hot encoding categorical columns¶

Define features (X) and target (y)¶

Split the data into training and testing sets¶

Standardize the numerical features¶

4. Exploratory Data Analysis (EDA)¶

Univariate analysis of numerical columns¶

Checking summary statistics¶

Univariate analysis for categorical variables¶

Printing the % sub categories of each category¶

Visualise Numerical Columns¶

Bivariate and Multivariate analysis¶

ML Models development using LR and SVM¶

Logistic Regression¶

Predictions for logistic regression¶

Evaluation Metrics for logistic regression¶

Confusion matrix for logistic regression¶

Support Vector Machine¶

Predictions for SVC¶

Evaluation Metrics for SVC¶

Evaluate the performance of the developed model using accuracy, precision, recall, F1 score and AUC-ROC¶

Hypertuning¶

Define the parameter grid to tune¶

Initialize GridSearchCV with 5-fold cross-validation¶

Fit the grid search¶

Best hyperparameters and model¶

Predictions and evaluation¶

Confusion matrix¶

Hyperparameter tuning for logistic regression¶

Perform Grid Search¶

Increase the number of iterations for Logistic Regression¶

Evaluate Performance Metrics for the Best Logistic Regression Model¶

Logistic Regression vs. SVM Comparison after tuning¶

Visualise Performance Metrics¶

Detailed Comparison¶

Summary Analysis¶