[BUG] zero_inflated_regressor.py

[jcoding2022]

The Bug

• The ZeroInflatedRegressor class consists of a classifier and a regressor and it says in the documentation that "The regressor is only trained on examples where the target is non-zero, which makes it easier for it to focus." In the code, however, the regressor is actually trained on examples where the classifier-predicted target is non-zero.

How to Fix the Bug

• Just replace line 111 with the following code:

non_zero_indices = np.where(y != 0)[0]

[FBruzzesi]

Oh gosh! Thanks for reporting this!
I don't think it was intended, but will double check and get back to you or fix it in the next release 👌