I am trying to implement a custom estimator class and use it with sklearn cross_val_score(). I have a "fit", "get_params", and "predict" functi