Predicting Customer Churn: An Analysis of Key Indicators and Retention Strategies

Customer churn is a critical problem for businesses as it can lead to a loss of revenue and customer loyalty. In this project, you will be able to identify the key indicators of customer churn and develop retention strategies to reduce customer attrition.

Methodology

To achieve your goal, follow these steps:

1. Data Collection: Collect data containing information on customers' demographics, services used, and payment history.

import pandas as pd
df_churn = pd.read_csv('churn_data.csv')        

2. Data Cleaning and EDA: Clean the data and conduct exploratory data analysis to identify patterns and trends within the data.

Univariate Analysis

# Create a histogram of MonthlyCharges

plt.hist(df_churn['MonthlyCharges'], bins=20)

plt.title('Monthly Charges')

plt.xlabel('Charges')

plt.ylabel('Frequency')

plt.show()        

Multivariate Analysis

#?Histogram?of?monthly?charges?and?churn
sns.histplot(data=df_churn,?x='MonthlyCharges',?hue='Churn',?multiple='stack',?kde=True)
plt.show()        

3. Hypothesis Development: Based on the EDA results, develop hypotheses to investigate the relationships between independent variables and customer churn.

Required libraries:

import pandas as pd

import numpy as np

import scipy.stats as stats        

4. Classification Algorithms: Build classification models to predict customer churn based on the independent variables.

# Importing necessary library
from sklearn.tree import DecisionTreeClassifier
# Splitting the data into train and test set
from sklearn.model_selection import train_test_split
# Defining the dependent and independent variables
X = df_churn.drop('Churn', axis=1)
y = df_churn['Churn']
# Splitting the dataset into training and test set
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Fitting the decision tree model
dt = DecisionTreeClassifier()
dt.fit(X_train, y_train)
# Predicting the target variable for the test set
y_pred = dt.predict(X_test)
# Evaluating the accuracy of the model
from sklearn.metrics import accuracy_score, confusion_matrix, classification_report
print('Accuracy:', accuracy_score(y_test, y_pred))
print('Confusion Matrix:\n', confusion_matrix(y_test, y_pred))
print('Classification Report:\n', classification_report(y_test, y_pred))

Accuracy: 0.7306325515280739
Confusion Matrix:
[[855 178]
[201 173]]
Classification Report:

???????????????precision? ? recall? f1-score ? support

?0.0 ? ? ? 0.81? ? ? 0.83? ? ? 0.82? ? ? 1033

?1.0 ? ? ? 0.49? ? ? 0.46? ? ? 0.48 ? ? ? 374
?accuracy ? ? ? ? ? ? ? ? ? ? ? ? ? 0.73? ? ? 1407
macro avg ? ? ? 0.65? ? ? 0.65? ? ? 0.65? ? ? 1407
weighted avg ? ? ? 0.73? ? ? 0.73? ? ? 0.73? ? ? 1407??        

5. Findings and Retention Strategies: Use the results of your analysis to develop retention strategies aimed at reducing customer churn and improving customer satisfaction.

from sklearn.model_selection import GridSearchCV
# Create an AdaBoostClassifier object
ada = AdaBoostClassifier()
# Define the hyperparameter grid
param_grid = {

????"n_estimators": [50, 100, 150],

????"learning_rate": [0.01, 0.1, 1.0]

}

# Use GridSearchCV to search for the best hyperparameter combination

grid_search = GridSearchCV(ada, param_grid=param_grid, cv=5)

grid_search.fit(X_train, y_train)

# Print the best hyperparameter combination and the corresponding accuracy score

print("Best hyperparameters: ", grid_search.best_params_)

print("Best accuracy score: ", grid_search.best_score_)

Best hyperparameters:? {'learning_rate': 1.0, 'n_estimators': 50}

Best accuracy score:? 0.7982222222222223        

Hypothesis

• Null hypothesis: There is no significant relationship between any of the independent variables (gender, senior citizen status, partner, dependents, tenure, phone service, multiple lines, internet service, online security, online backup, device protection, tech support, streaming TV, streaming movies, contract, paperless billing, payment method, monthly charges, total charges) and customer churn.

• Alternative hypothesis: There is a significant relationship between at least one of the independent variables and customer churn.

Questions

1. Is there a correlation between contract length and customer churn?

# Calculate correlation coefficient

corr_coef = df_churn['Contract'].corr(df_churn['Churn'])

# Create scatterplot with regression line

sns.lmplot(x='Contract', y='Churn', data=df_churn)

# Add correlation coefficient to plot

plt.title(f"Correlation between Contract and Churn: {corr_coef:.2f}")

plt.show()        

2. Do customers who have online security and backup services have lower churn rates?

# create a countplot to compare churn rates for customers with and without online security

sns.countplot(x="OnlineSecurity", hue="Churn", data=df_churn)

# create a countplot to compare churn rates for customers with and without backup services

sns.countplot(x="OnlineBackup", hue="Churn", data=df_churn)        

3. Does the payment method have an impact on customer churn?

# create a countplot to compare churn rates for different payment methods

sns.countplot(x="PaymentMethod", hue="Churn", data=df_churn)        

4. Is there a difference in churn rates between male and female customers?

# create a countplot to compare churn rates for male and female customers

sns.countplot(x="gender", hue="Churn", data=df_churn)        

5. Are customers with dependents less likely to churn compared to those without dependents?

# create a countplot to compare churn rates for customers with and without dependents

sns.countplot(x="Dependents", hue="Churn", data=df_churn)        

Results and Conclusion

Based on your analysis, here are possible findings:

• Contract length has a significant impact on customer churn, with customers on month-to-month contracts having higher churn rates.
• Customers who have online security and backup services have lower churn rates.
• The payment method has an impact on customer churn, with customers using electronic checks having higher churn rates.
• There is no significant difference in churn rates between male and female customers.
• Customers with dependents are less likely to churn compared to those without dependents.

To reduce customer churn and improve customer satisfaction, you can recommend the following strategies:

• Encouraging customers to switch to longer-term contracts
• Offering online security and backup services to customers
• Encouraging customers to switch to alternative payment methods such as credit cards
• Targeting marketing efforts towards customers with dependents

By identifying key indicators of customer churn and developing retention strategies, your business can reduce customer attrition and improve customer satisfaction. We encourage you to conduct similar analyses to optimize your customer retention efforts.

Contact us today to learn more about our data analysis and machine learning services and how we can help you improve customer retention and boost revenue.