Data scientists and demand planners facing a new forecasting problem need expert guidance on which forecasting methodology to use — from classical ARIMA and exponential smoothing to gradient boosting regressors and transformer-based neural forecasters — given their data characteristics and operational constraints.