← Back to Portfolio
Customer Churn Prediction with What-If Scenarios
Product AnalyticsML-powered churn prediction model with interactive scenario planning for subscription businesses.
Overview
💡 Challenge
Subscription businesses need to identify at-risk customers early and understand which interventions are most effective at preventing churn.
⚡ Solution
Built a Random Forest classification model to predict churn probability, integrated with a Streamlit interface allowing users to test "what-if" scenarios (e.g., discounts, feature access).
🎯 Impact
Enables customer success teams to proactively target high-risk accounts and quantify the expected impact of retention strategies.
Technical Details
🛠️ Tech Stack
PythonScikit-learnStreamlitPandasPlotly
✨ Key Features
- ✓Churn probability scoring for individual customers
- ✓Feature importance visualization
- ✓Interactive what-if scenario testing
- ✓Segment-level churn analysis
Key Learnings
- →Class imbalance requires SMOTE or weighted models
- →Feature engineering from behavioral data is critical
- →Stakeholders value interpretability over marginal accuracy gains
📊 Data Notes
This project uses synthetic/open data to demonstrate capabilities while maintaining privacy and confidentiality. All methods and approaches are applicable to real-world scenarios.