CODEWITHSUNDEEP
scikit-learngrid-search
AMisra
AMisra

Reputation: 1889

Using GridSearchCV gives an error with CalibratedClassifierCV in scikit-learn. GridSearchCV object has no attribute 'best_params_

I use CalibratedClassifierCV with RandomForest and use GridSearch to determine the best parameters. However, when I use GridSearchCV to read back the best parameters it says GridSearchCV object has no attribute 'best_params_'

from sklearn.calibration import CalibratedClassifierCV
from classifiers import SVMClassification 
from sklearn.model_selection import  GridSearchCV
from imblearn.pipeline import Pipeline as imbpipeline
        
pipeline=imbpipeline([ ('oversample', Sampling(random_state=444)),('rf', rf())])
paramgrid=[ {'rf__max_depth': [8,10], 'rf__n_estimators':[3,5]}]           
grid_search_rf = GridSearchCV(pipeline, param_grid=paramgrid,cv=3)
rf_calibrated=CalibratedClassifierCV(grid_search_rf,cv=5, method="sigmoid")
rf_calibrated.fit(x_labelled,y_labelled)
print(rf_calibrated.base_estimator.best_params_)

Output

AttributeError: 'GridSearchCV' object has no attribute 'best_params_'

Upvotes: 1

Views: 2088

Answers (1)

Vivek Kumar
Vivek Kumar

Reputation: 36599

I am assuming you are thinking that the CalibratedClassifierCV will fit the supplied estimator and then somehow enhance (calibrate) the output probabilities for that.

Thats partially correct.

What happens is:

  • CalibratedClassifierCV will make a clone of the supplied estimator and then fit the data onto the clone. So you doing this 

    rf_calibrated.base_estimator`

    will only return an unfitted estimator which don't have the best_params_ attribute. best_params_ is only available after fitting.

Checking best_params_ from CalibratedClassifierCV will not make any sense, because it will divide the data into 5 parts (as you have done cv=5) in it and each fold is trained on a separate clone, so you may have multiple best_params in that depending on the data.

Upvotes: 2

Related Questions