python - Sklearn GridSearchCV, class_weight not working for unknown reason :( -

trying class_weight going . know rest of code works, class_weight gives me error:

    parameters_to_tune = ['min_samples_split':[2,4,6,10,15,25], 'min_samples_leaf':[1,2,4,10],'max_depth':[none,4,10,15],                                              ^ syntaxerror: invalid syntax 

here code

clf1 = tree.decisiontreeclassifier()  parameters_to_tune = ['min_samples_split':[2,4,6,10,15,25], 'min_samples_leaf':[1,2,4,10],'max_depth':[none,4,10,15],  'splitter' : ('best','random'),'max_features':[none,2,4,6,8,10,12,14],'class_weight':{1:10}] clf=grid_search.gridsearchcv(clf1,parameters_to_tune) clf.fit(features,labels) print clf.best_params_ 

does spot mistake making ?

i assume want grid search on different class_weight 'salary' class.

the value of class_weight should list:

'class_weight':[{'salary':1}, {'salary':2}, {'salary':4}, {'salary':6}, {'salary':10}] 

and can simplify list comprehension:

'class_weight':[{'salary': w} w in [1, 2, 4, 6, 10]] 

the first problem parameter values in dict parameters_to_tune should list, while passed dict. can fixed passing list of dicts value of class_weight instead , each dict contains set of class_weight decisiontreeclassifier.

but more serious problem class_weight weights associated classes, in case, 'salary' name of feature. can not assign weights features. misunderstood intention @ first confused want.

the form of class_weight {class_label: weight}, if mean set class_weight in case, class_label should values 0.0, 1.0 etc., , syntax like:

'class_weight':[{0: w} w in [1, 2, 4, 6, 10]] 

if weight class large, more classifier predict data in class. 1 typical case use class_weight when data unbalanced.

here example, although classifier svm.

update:

the full parameters_to_tune should like:

parameters_to_tune = {'min_samples_split': [2, 4, 6, 10, 15, 25],                       'min_samples_leaf': [1, 2, 4, 10],                       'max_depth': [none, 4, 10, 15],                       'splitter' : ('best', 'random'),                       'max_features':[none, 2, 4, 6, 8, 10, 12, 14],                       'class_weight':[{0: w} w in [1, 2, 4, 6, 10]]}