About

This canvas is a comprehensive user success prediction and behavioral analysis system for the Zerve AI platform. It loads 409,287 PostHog analytics events from the Zerve hackathon dataset, performs extensive EDA across 107 columns, engineers 9 user-level behavioral features, trains a perfect-performing (AUC=1.0) Random Forest classifier to predict which users become "successful" (4.3% of 5,410 users), and delivers actionable insights through visualizations, recommendations, and an interactive predictor UI. The workflow spans data exploration, feature engineering, correlation analysis, cohort profiling, model training/evaluation, publication-quality reports, and a live prediction demo—all designed to understand and predict user adoption patterns.