Exercise

Analyzing the best results

At the end of the day, we primarily care about the best performing 'square' in a grid search. Luckily Scikit Learn's gridSearchCv objects have a number of parameters that provide key information on just the best square (or row in cv_results_).

Three properties you will explore are:

  • best_score_ – The score (here ROC_AUC) from the best-performing square.
  • best_index_ – The index of the row in cv_results_ containing information on the best-performing square.
  • best_params_ – A dictionary of the parameters that gave the best score, for example 'max_depth': 10

The grid search object grid_rf_class is available.

A dataframe (cv_results_df) has been created from the cv_results_ for you on line 6. This will help you index into the results.

Instructions

100 XP
  • Extract and print out the ROC_AUC score from the best performing square in grid_rf_class.
  • Create a variable from the best-performing row by indexing into cv_results_df.
  • Create a variable, best_n_estimators by extracting the n_estimators parameter from the best-performing square in grid_rf_class and print it out.