ValueError: max_features must be in (0, n_features]

[imb-ai-junghwan]

Hi, thanks for your great achievement.
I got an error when i try to run this model.

I used learner type is rf and feature selection is lasso.
but, i got an error 'ValueError: max_features must be in (0, n_features]' in classification_thomas-manghi.py
I think this error occurred in below block.
"if lp.learner_type == 'rf':
for k in steps:
clf_f = []
p_es_f = []
l_es_f = []

			if lp.refine == 'rf':	
				for j in range(runs_n*runs_cv_folds):
					clf_f.append(\
						RandomForestClassifier(n_estimators=par['number_of_trees']
						, criterion=par['rf_criterion']\
						, **max_features=(k if par['feature_identifier'] != 'UniRef90' else (k if k<=128 else 0.3))**" .

My argument of feature_identifier is default. so, i don't know how to fix it for running without error.

Can you give me some advice or suggestions?

Thanks.

[edoardopasolli]

In your input file, are your features/taxas identified by the prefix "k__" which is the default for the parameter feature_identifier ?

[imb-ai-junghwan]

Sure. my features/taxa has "k__" prefix.

[imb-ai-junghwan]

Can i change the 'max_features' value to '(k if k < n_features else n_features)' ?

I'm asking because I'm not sure if this is a change that will harm the your intention.