Back to Projects
🎯 Targeting & Growth
Predicting Churn & Design of Experiments
UCSD · Customer Analytics
Built predictive churn models and designed controlled experiments to test which retention interventions actually work — moving beyond correlation to measure causal impact of specific tactics on at-risk customer segments.
Approach
Developed churn classifiers using behavioral and transactional features, then designed A/B testing frameworks to validate proposed retention strategies before committing to full rollout. The experimental design ensured statistical rigor in measuring treatment effects.
Tools & Technologies
PythonScikit-learnStatsModelsA/B Testing Frameworks
Topics
Churn PredictionA/B TestingExperiment DesignPython