Executive Summary

Project Overview

The Customer Churn Analyzer is a data-driven solution designed to identify at-risk customers in telecom and subscription-based business models. By leveraging machine learning, the tool analyzes historical customer behavior—ranging from billing patterns to support interactions—to predict the likelihood of service cancellation (churn).

This project provides a complete end-to-end pipeline, from exploratory data analysis (EDA) and feature engineering to a predictive Logistic Regression model, enabling businesses to shift from reactive to proactive retention strategies.

Key Insights & Findings

Analysis of the customer dataset (1,002 records) revealed several critical indicators that correlate strongly with customer attrition:

• Support Intensity: High frequencies of support calls are the strongest predictor of churn. Customers experiencing recurring technical or service issues are significantly more likely to leave.
• Contractual Sensitivity: Customers on "Monthly" contracts exhibit much higher churn rates compared to those on long-term commitments.
• Financial Friction: Higher monthly bills and the absence of AutoPay features contribute to higher turnover, suggesting that billing friction and price sensitivity are key pain points.
• Tenure Stability: New customers (lower tenure months) are at the highest risk, while churn rates stabilize as customer loyalty matures over time.

Predictive Capabilities

The current implementation utilizes a Logistic Regression classifier. While providing a solid baseline for understanding feature influence, the model achieves a moderate accuracy of approximately 54%, indicating that churn behavior in this dataset is complex and may require more advanced ensemble methods for high-precision targeting.

Example Usage

The system allows for real-time prediction by passing customer attributes into the trained model:

# Example: Predicting churn for a high-risk customer profile
new_customer = pd.DataFrame({
    'Contract': [0],          # 0: Monthly
    'SupportCalls': [6],      # High volume of calls
    'MonthlyBill': [110.0],   # Premium billing
    'PaymentMethod': [2],     # Credit Card
    'BillingIssues': [1],     # Existing issues
    'DataUsageGB': [85.0],
    'TenureMonths': [3],      # New customer
    'AutoPay': [0]            # Manual billing
})

prediction = model.predict(new_customer)
# Output: Predicted Churn: Yes

Strategic Recommendations

Based on the data analysis, the following actions are recommended to improve customer retention:

1. Proactive Support Outreach: Implement a "Red Flag" system for customers who exceed 4 support calls in a single billing cycle.
2. Incentivize Long-term Contracts: Offer discounts or value-added services to migrate "Monthly" users to annual plans.
3. AutoPay Promotion: Reduce billing-related churn by offering small incentives (e.g., 5% discount) for enabling automated payments.
4. Loyalty Programs: Focus retention marketing on customers within their first 6 months of service, as this is the highest risk window.

Future Roadmap

To increase the precision of the analyzer, future iterations will focus on:

• Integrating Random Forest or XGBoost classifiers to capture non-linear relationships.
• Implementing SHAP (SHapley Additive exPlanations) to provide transparent, per-customer reasons for churn predictions.
• Developing a web-based dashboard for non-technical stakeholders to upload CSV files and receive batch predictions.