I built an airline loyalty churn prediction and retention system that processed 16,737 records, selected its model using PR-AUC, and more than doubled churner recall compared with the baseline. I also segmented customers into behavioral cohorts and shipped a Streamlit dashboard to help the business team use churn risk in retention planning.
For Rossmann Store Sales, I built a forecasting pipeline and dashboard that compared LSTM with XGBoost; the tuned XGBoost model reached an R² of 0.928. I also created a Power BI loan analytics dashboard covering 38.6K applications, with DAX measures and interactive views for portfolio and credit-risk analysis.
