End-to-end ML pipeline across 45 stores — five models benchmarked, Optuna-tuned, a 4-page Power BI report, and a 5-page Streamlit app. Best result: 2.17% MAPE with Prophet.
What I built
Model benchmark
Evaluation window: May – Oct 2012 · 20% holdout · temporal split · no data leakage · Prophet + regressors (5th model) in the repo
Feature importance
Forecast
Illustrative shape — the real forecast plots are in the live app.
Streamlit app — full interactive forecasting dashboard, model comparison, and future projections.
Built by SKAY · Oussama Skia — AI / ML Engineer · Casablanca