Predictability Check: Demand Forecasting
summary
Step-by-step Python pipeline that engineers lag/shift features, rolling-window stats and holiday flags, trains XGBoost models and exports a simple Streamlit app for interactive per-customer–per-product forecasts. Claims: overall test RMSE ≈ 1.28 units and R² ≈ 0.60 across combos; top predictors are product_id and a 40-day rolling statistic labeled window40. Code uses pandas for preprocessing, label encoding for categories, XGBoost for training, and a dashboard that shows predicted vs actual series and per-combination error metrics.
AI-written summary · about
chapters (14)
- 0:00Demand forecasting intro
- 4:38Why accurate forecasting matters
- 6:31Time series predictability
- 11:55Validate forecasting data
- 12:19Food industry case studies
- 15:11ML forecasting benefits
- 16:17Python forecasting model
- 28:22Python setup & data exploration
- 33:27Feature engineering for sales
- 35:37Train XGBoost model
- 40:42Model performance metrics
- 46:12Streamlit dashboard results
- 52:29Model limits & improvements
- 56:02Conclusion & Q&A
related talks
28mDemand Forecasting: Accuracy, Stability, Automation
40mMulti-Series Forecasting at Scale with StatsForecast
31mFrom stockouts to happy customers: Proven solutions for time series forecasting in retail
09mDemand Forecasting in Supply Chain Using AI Basis Machine Learning
42mRobust Automated Forecasting in Python and R
28mMastering Demand Forecasting: Lessons from Europe's Largest Retailer
29mProbabilistic demand forecasting at scale
01h 23mA Hands-On Introduction to Your First Data Science Project