Customer Retention Intelligence System

A machine learning–driven customer churn prediction and retention analytics platform designed to help subscription businesses identify at-risk customers and take proactive action.

🚀 Overview

This project builds a complete end-to-end churn intelligence system using real telecom customer data. It combines exploratory data analysis, statistical insights, and machine learning to predict customer churn and quantify revenue at risk.

📊 Dataset

IBM Telco Customer Churn Dataset (7,000+ customers)

Features include:

• Tenure
• Monthly & Total Charges
• Contract Type
• Internet Service
• Payment Method
• Customer Demographics

Target:

• Churn (Yes / No)

🔍 Key Insights

• Month-to-month customers churn 15× more than two-year contract users
• Fiber optic users show the highest churn
• High-paying customers are more likely to leave

🤖 Machine Learning

Model: Logistic Regression
Metrics:

• ROC AUC ≈ 0.80+

The model predicts churn probability for each customer and identifies the strongest churn drivers.

🧠 Churn Intelligence Engine

The system generates:

• Individual churn risk scores
• High-risk customer lists
• Revenue at risk estimates

This enables targeted retention strategies such as discounts, service improvements, and contract upgrades.

📁 Project Structure

Customer-Retention-Intelligence-System/
├── data/
├── notebooks/
├── models/
├── reports/
└── README.md

🛠 Tech Stack

• Python
• Pandas, NumPy
• Scikit-learn
• Matplotlib

📌 Use Case

Designed for subscription-based businesses such as telecom, SaaS, streaming, and e-commerce platforms to improve customer retention and reduce revenue loss.