GridSearchCV for specificity?

Viewed 161

Wanted to do grid search for recall and specificity but could not find specificity as a valid scoring parameter. I went to the scoring parameters for sklearn and did not see it in the list of valid parameters. Is there a work around or something I am missing in the parameter list?

This is what I have so far:

# Just some random example model. 
pipe_rf = make_pipeline(RandomForestClassifier(criterion='gini',
                                               n_estimators=25, 
                                               max_features=2, 
                                               max_depth=3,
                                               random_state=1))

# Just filled in some parameters. 
param_grid = [{'randomforestclassifier__n_estimators': [1, 2, 5, 10, 25, 50, 100],
              'randomforestclassifier__max_features': [1, 2, 5, 10, 25, 50, 100],
              'randomforestclassifier__max_depth': [1, 2, 5, 10, 25, 50, 100]}]

##### Perform the grid search #####
# *** Want to include specificity as a scoring metric *** #
gs = GridSearchCV(estimator=pipe_rf, 
                  param_grid=param_grid, 
                  scoring='recall', 
                  refit=True,
                  cv=10,
                  n_jobs=-1)
0 Answers
Related