Extracting scores from gridsearch object

extract_results_gridsearch.py

def extract_results_GS(grid_scores):
    """
    After applying gridsearch to a dataset, for all different combinations of parameters, an array of scores is
    given in the grid_scores object of the estimator. The length of this array is equal to the number of fits of 
    the model. E.g. a model that is fit on three different train/test combinations, as happens in cross-validation. From that, the 
    mean validation score and standard deviation of the scores can be obtained. This function converts a 
    GridSearchCV.grid_scores_ object to a pandas DataFrame object. The frame is unsorted, because the best parameter
    setting can be verified in the GridSearchCV.best_estimator_ and best_score_ properties.
    """
    
    import numpy as np
    import pandas as pd
    
    scores = [item.mean_validation_score for item in grid_scores]
    stddevs = [item.cv_validation_scores.std() for item in grid_scores]
    parameter_sets = [item.parameters for item in grid_scores]

    return pd.concat([pd.DataFrame(parameter_sets, index=np.arange(len(grid_scores))), pd.Series(scores, name='mean_val_score'), 
              pd.Series(stddevs, name='stddevs_val_score')], axis=1)