About

This canvas implements an end-to-end telecommunications churn prediction pipeline: it loads and cleans the IBM Telco Customer Churn dataset (7,032 customers), performs exploratory analysis with 5 publication-ready visualizations, and trains a multi-model comparison (Gradient Boosting, Random Forest, Logistic Regression, SVM) ultimately selecting Random Forest as the best performer (CV AUC 0.847) with detailed evaluation charts and business insights.