About

This project analyzes user behavior data from the Zerve platform to identify which workflows and engagement patterns are most predictive of long-term user success. Using event-level logs across notebooks, runs, and deployments, I aggregated activity into user-level behavioral features such as active days, early engagement, consistency, and usage intensity.

Long-term success was defined as sustained activity for at least 30 days after a user’s first interaction. Comparative analysis and an interpretable logistic regression model were used to identify key drivers of retention. The analysis reveals that early engagement and consistent activity are stronger predictors of success than raw usage volume. These insights can help improve onboarding, product design, and user retention strategies.