All projects

CreatorPulse

Creator-economy burnout prediction with causal and survival analysis.

Problem

Platforms want to retain creators, but churn and burnout are driven by causal factors that simple correlation models miss. Understanding what actually moves the needle requires rigorous statistical method, not just prediction.

Solution

On synthetic creator data, a predictive model is paired with causal inference and survival analysis to estimate not just who will burn out, but why and when, validated with A/B testing methodology.

Architecture

An XGBoost model predicts burnout risk, difference-in-differences causal inference estimates treatment effects, survival analysis models time-to-burnout, and A/B testing frames the experimental design, all on synthetic data.

Tech stack

XGBoostCausal Inference (DiD)Survival AnalysisA/B TestingPython