QM020 Hands-on Lab
Time-Series Model Selection and Hyperparameter Tuning
Objective
Tune a Ridge penalty using expanding chronological folds, then evaluate the selected procedure on a later holdout block.
Run
python python/qm020_time_series_tuning.pyExperiment
Change gap, the minimum training size, or the strength of the synthetic regime change. Explain why a gap is justified or unnecessary from the label-information design rather than choosing it mechanically.
The data are synthetic and provide no empirical forecasting claim.