Evaluate temporal adaptation with additional machine-learning predictors

Evaluate the behavior of the temporal adaptation policy with machine-learning predictors beyond XGBoost and Random Forest.

Background

The paper evaluates its Core + Additive temporal adaptation policy primarily with XGBoost and, in an ablation study, with Random Forest. Because the policy may behave differently with other predictor families, the authors identify evaluation with additional machine-learning predictors as an unresolved area for future work. The problem is specifically to assess whether the observed temporal-adaptation behavior generalizes beyond the two predictors studied.