Two forecasting case studies: demand for slow-moving spare parts on an industry client project, and a comparison of forecasting models across two book sales datasets.
An employer project forecasting slow-moving spare parts, then generalised across all four demand classes and tested on two public datasets. Why the accuracy figures mislead, and where a forecast stops being worth running.
Compared ARIMA, SARIMA, XGBoost, LSTM and hybrid models to forecast book demand across two datasets.