A/B test analyzer that returns ship/hold/iterate/kill, not a p-value. Power vs a pre-specified MDE, CIs, effect size, and SRM checks on every result. 37 tests.

ab-testingexperimentationhypothesis-testingproduct-analytics

python

CCallahan308

Repo · 1 stars

▲ 0

Profile README

churn-roi-simulator

Repo · 2 stars · Python

▲ 0

Event-log churn scoring turned into a retention-budget decision: leakage-safe time-windowed labeling, calibrated LightGBM vs a logistic baseline, ROI simulator on 2.76M RetailRocket events. Honest result: baseline wins the holdout; CV 0.88 ± 0.06.

churn-predictionlightgbmmachine-learningmlflowpydantic

ecommerce-retention-growth

Repo · 2 stars · Python

▲ 0

Churn, LTV segmentation, and win-back ROI on the KKBox/WSDM dataset: calibrated XGBoost (ROC-AUC 0.788 on a fixed-seed 50K-user sample), K-Means LTV segments, retention-budget simulator with synthetic cold-start mode.

churn-predictioncustomer-segmentationkkboxmachine-learningmkdocs

ny-sparcs-healthcare-bi

Repo · 1 stars · Python

▲ 0

Healthcare operations BI on NY SPARCS inpatient data: DuckDB + SQL marts + Streamlit, case-mix adjusted facility benchmarks and XmR process control

signalforge

Repo · 1 stars · Python

▲ 0

Production churn prediction with statistical rigor — Optuna-tuned models, bootstrap 95% CIs, paired t-tests, calibration analysis on IBM Telco data. Live Streamlit dashboard.

bootstrapcalibrationchurn-predictionconfidence-intervalsmachine-learning