Actually @jnothman @RahulDamineni @alexsieusahai I don't think we should provide a parameter for the norm to make coef_ values non-negative, RFE and RFECV do not. We should use the exact same code as what is there, at least as a first implementation:

scikit-learn/sklearn/feature_selection/rfe.py

Lines 181 to 195 in bac89c2

    
           # Get coefs 
        
           if hasattr(estimator, 'coef_'): 
        
               coefs = estimator.coef_ 
        
           else: 
        
               coefs = getattr(estimator, 'feature_importances_', None) 
        
           if coefs is None: 
        
               raise RuntimeError('The classifier does not expose ' 
        
                                  '"coef_" or "feature_importances_" ' 
        
                                  'attributes') 
        
           # Get ranks 
        
           if coefs.ndim > 1: 
        
               ranks = np.argsort(safe_sqr(coefs).sum(axis=0)) 
        
           else: 
        
               ranks = np.argsort(safe_sqr(coefs))

	# Get coefs
	if hasattr(estimator, 'coef_'):
	coefs = estimator.coef_
	else:
	coefs = getattr(estimator, 'feature_importances_', None)
	if coefs is None:
	raise RuntimeError('The classifier does not expose '
	'"coef_" or "feature_importances_" '
	'attributes')

	# Get ranks
	if coefs.ndim > 1:
	ranks = np.argsort(safe_sqr(coefs).sum(axis=0))
	else:
	ranks = np.argsort(safe_sqr(coefs))

Uh oh!

AdaBoost feature_importances_ should support base estimators with coef_ #12137

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions