Tuning classification model on different metrics (lift, recall, precision,...)

[hendrikdewinter8]

Tuning classification model on different metrics (lift, recall, precision,...)

now a classification cobra model can only be tuned on AUC

Task Description

This issue will...

[sborms]

This can already be done with compute_model_performances() and its metric argument, but by default for logistic regression higher = better (~AUC) and for linear regression lower = better (~RMSE), so needs to be documented more clearly (or, as in PyCaret I think, adding an option to say higher_is_better=True or alike).