retail-forecasting

Retail-Forecast — tools, notebooks, and models for demand forecasting and time-series analysis 📈🧠 Designed for retail use-cases (sales, promotions, inventory) with reproducible pipelines and experiments 🛒🔁

M5-inspired retail demand forecasting in Python. A custom evaluation harness compares LightGBM, AutoARIMA and seasonal naive, selects the winner, and drives inventory recommendations and drift monitoring. Built in under 7 days.

Production-style retail sales forecasting project with leakage-safe time-based validation, LightGBM, CatBoost, feature engineering, inference pipeline, and Streamlit dashboard.

Retail demand forecasting using statistical time series and machine learning models, with feature engineering and model comparison.

Multi-model time series forecasting platform: SARIMA, LightGBM (SMAPE 12%), and N-BEATS (PyTorch, from scratch). Walk-forward validation across 8 expanding windows. Lag/rolling/Fourier feature engineering. 16 tests.

End-to-end retail sales forecasting using LightGBM with time-series features, SHAP explainability, FastAPI inference, Streamlit demo, and CI for production-ready ML workflows.

Retail sales forecasting with expanding-window rolling-origin validation, direct XGBoost, and an integrated analytical report. Includes synthetic-data variant of the original submission.

Code and frozen evidence for retail forecast-replenishment evaluation under information and policy uncertainty.

End-to-end Machine Learning, Deep Learning, SQL, and NLP capstone projects (Projects 01 to 07) completed during the Labmentix AI & Data Science Internship.