This project is an end-to-end data science solution designed to optimize retail operations. It transforms 73,000+ raw transactions into a strategic decision-making engine using a dual-layered AI approach: identifying high-value store segments and forecasting revenue trends.
- Analytics Engine: Python (Pandas, Scikit-Learn, XGBoost)
- Business Intelligence: Power BI (Interactive Executive Dashboards)
- Clustering: K-Means for Store Tiering
- Forecasting: XGBoost Regression
- The Promo Paradox: Statistical T-Testing revealed a p-value of 0.68, proving that broad promotions were not significantly driving revenue lift—leading to a recommendation to pivot to Tier-specific premium experiences.
- Predictive Accuracy: The XGBoost model achieved a Mean Absolute Error (MAE) of $4,718, providing a high-confidence window for inventory and labor scheduling.
- Strategic Segmentation: Successfully clustered 100+ locations into Flagship (Tier 1), Growth (Tier 2), and Value (Tier 3) stores for targeted marketing spend.
Retail_Forecasting_Engine.ipynb: Full Python pipeline for cleaning, clustering, and modeling.Retail_Intelligence_Dashboard.pbix: Interactive Power BI report (Requires Power BI Desktop).Final_Enterprise_Master_Data.csv: The "Golden Dataset" engineered for the final model.Executive_Summary.pdf: Formal business case and strategic recommendations.
- Model: Run the Jupyter Notebook to see the data engineering and model validation steps.
- Dashboard: Open the
.pbixfile to interact with the revenue heatmaps and price elasticity charts.