About

Comprehensive User Retention & Success Prediction Pipeline — A complete data science workflow analyzing 141k+ events from 8,000+ users to predict product success. The canvas ingests raw retention data, performs multi-stage exploratory analysis (event taxonomy, cohort retention, feature adoption), engineers 34 behavioral features across 6 pillars (retention, depth, breadth, growth, collaboration/deployment), trains 4 ML models (LR, RF, XGB, GB) to classify successful vs. regular users, applies SHAP/TreeExplainer for feature importance ranking, creates 5-pillar success score (0-1 normalized), and segments users into 5 data-driven personas (Power Builders, Collaborative Builders, Rising Stars, Window Shoppers, Dormant Accounts, Deep Divers, Steady Explorers) with product recommendations per segment.