Make the explain module independent of the regressor.

[HellenNamulinda]

Making the explain module independent of the regressor will alllow generating interpretability plots of any model, including classification models.

Also, use flaml to only suggest hyperparameters.