Instead, plot its components, and check that they make sense Many data scientists approach modelling as a feature engineering task: they gather features, do cross-validation, and if the score isn’t good enough, they keep adding more features until it is. In forecasting, where there is typically little data, this is…