Customer Churn Prediction
ML system predicting which customers will leave, with AUC of 0.8438 using ensemble models.
Project Overview
A machine learning pipeline trained on 7,000 telecom customer records to predict churn probability. Trains 4 models simultaneously — Logistic Regression, Random Forest, XGBoost, and Gradient Boosting — with full model comparison and feature importance analysis.
⚠ The Problem
Telecom companies lose significant revenue to customer churn. Identifying at-risk customers early allows retention teams to intervene before losing them.
✓ The Solution
Complete preprocessing pipeline with encoding and scaling, 4-model training comparison, feature importance visualization showing key churn drivers. Best model: Gradient Boosting AUC 0.8438.
Results & Outcomes
Best AUC 0.8438. Feature importance reveals top churn predictors. Predict single customer churn probability in real time.