About

This canvas is a comprehensive machine learning pipeline that analyzes Zerve user behavior to predict success—identifying which users become long-term, productive platform users through behavioral features like engagement scoring, funnel completion, and time-to-first-action. The workflow ingests event logs, engineers 16 behavioral features (session depth, high-value action counts, funnel progression flags), trains a gradient-boosted classifier (ROC-AUC ~0.81), visualizes feature importance and funnel drops, and surfaces actionable product insights (e.g., "Deploy is the highest-signal action" and "Time-to-first-action is the #1 predictor").