Customer Churn Prediction
Predicting which customers are likely to churn for a SaaS platform.
Overview
Reducing churn directly increases ARR and customer lifetime value.
Problem & Constraints
High churn rates impact revenue and growth for SaaS businesses.
Data
100k+ user records, tabular, time-series, engineered features, no leakage.
Approach
Baseline: Logistic Regression → XGBoost → Feature selection → SHAP analysis.
Evaluation
AUC: 0.89, Precision@Top10%: 0.72
Results
Reduced churn by 18% in pilot, actionable insights for retention.
Production / Engineering
- Inference design: See repo for details
- Latency & cost: See highlights
- Monitoring: Monitoring ideas in repo
Demo
Demo coming soon.
Links
What I'd Do Next
See repo for future improvements and roadmap.