A ferry-delay model is weak on both the training slice and the validation slice. Someone adds heavier dropout and a stronger L2 penalty. What should change instead?
Select an answer to reveal the explanation.
Short Explanation
Weak on both the training slice and the validation slice means the model is too small. Raise capacity with more trees, more depth, or a richer family. Heavier dropout and L2 shrink an already-weak fit, Polly does not grow a model, and cutting trees is the wrong way.
Full Explanation
Weak performance on both train and validation is underfitting: capacity is too low or the algorithm is too simple. Increase trees, depth, or model family. Heavier dropout and L2 make underfitting worse. Polly is irrelevant, and cutting capacity is the wrong direction.