About

This project presents a fully reproducible, end-to-end analytics and machine learning workflow built entirely on Zerve.ai to understand and predict long-term user success on the Zerve platform.

Using over 409,000 real user interaction events across 4,700+ users, the project transforms raw event logs into structured behavioral insights that explain which user behaviors and workflows most strongly predict sustained engagement and success.

Rather than focusing on isolated actions, the analysis emphasizes workflow depth, iteration patterns, return consistency, and feature adoption, aligning closely with how real users derive long-term value from data platforms

What This Project Does
Aggregates raw event-level data into high-resolution user behavior profiles
Engineers 38+ behavioral features capturing engagement, reuse, and workflow sophistication
Discovers distinct user archetypes through behavioral pattern analysis
Trains interpretable machine learning models to predict long-term success
Produces actionable insights to guide onboarding, retention, and product decisions

🧪 Methodology Highlights
Data Processing: 409K events → 4.8K user-level feature vectors
Feature Engineering: Engagement metrics, workflow depth, reuse intensity, return consistency
Modeling: Logistic Regression, Decision Tree, Random Forest
Evaluation: AUC up to 0.986, with strong interpretability
Explainability: Feature coefficients, tree importance, and permutation importance
Statistical Rigor: Significance testing and effect size analysis
All analysis runs serverlessly inside Zerve, with no external infrastructure or datasets.

📊 Key Findings
Return consistency is the strongest predictor of long-term success
Users who adopt agent tools and advanced workflows early are significantly more likely to succeed
Workflow reuse and iteration matter more than raw activity volume
Clear separation exists between power users, explorers, steady contributors, and one-time visitors

🚀 Why This Matters
The insights from this project enable:
Early identification of at-risk users
Targeted onboarding interventions
Smarter feature prioritization
Data-driven customer success strategies