And I'm wanting to know how I could potentially implement this in Feature Selector. Instead of using eval_metric='auc', is there a way I can rig this to use my metric? I'm not super clear on how LightGBM implements custom scoring metrics, but looks confusing from first glance lol.
Not really an issue, just a general question, but I have made a scoring metric using the fbeta score that looks like this:
my_custom_scorer = make_scorer(fbeta_score, average='binary', zero_division=0, pos_label=1, beta=fbeta_for_scorer, greater_is_better=True)And I'm wanting to know how I could potentially implement this in Feature Selector. Instead of using
eval_metric='auc', is there a way I can rig this to use my metric? I'm not super clear on how LightGBM implements custom scoring metrics, but looks confusing from first glance lol.Thanks!