replace concat in encoders by y_train.groupby(X_train["variable"]).mean().sort_values()

[solegalli]

Applies to the following encoders:

• OrdinalEncoder
• MeanEncoder
• PRatioEncoder
• WoEEncoder

Currently we concatenate the target to the predictors (X_train) to determine the mappings. This is not necessary. We can determined them automatically with syntax like this:

y_train.groupby(X_train["A7"]).mean().sort_values()

therefore removing one unnecessary step in the calculation, the pd.concat step

[solegalli]

@noahjgreen295 @bmreiniger would this be an interesting issue to work on?