About

This canvas builds an end-to-end **Telco Customer Churn Prediction System** that downloads the Kaggle dataset, performs comprehensive EDA with custom visualizations, engineers 30+ features, and trains a tuned logistic regression model (ROC-AUC 0.83+) to identify high-risk churners with actionable retention strategies. The workflow spans data cleaning, exploratory analysis (distributions, outliers, correlations, categorical breakdowns), train-test split with scaling, hyperparameter tuning across three models, final evaluation with confusion matrices and PR curves, and real-world churn predictions on new customers with risk tier segmentation and business recommendations.