predictor.py 文件源码

def print_training_summary(self, gs):
        print('The best CV score from GridSearchCV (by default averaging across k-fold CV) for ' + self.output_column + ' is:')
        if self.took_log_of_y:
            print('    Note that this score is calculated using the natural logs of the y values.')
        print(gs.best_score_)
        print('The best params were')

        # Remove 'final_model__model' from what we print- it's redundant with model name, and is difficult to read quickly in a list since it's a python object.
        if 'model' in gs.best_params_:
            printing_copy = {}
            for k, v in gs.best_params_.items():
                if k != 'model':
                    printing_copy[k] = v
                else:
                    printing_copy[k] = utils_models.get_name_from_model(v)
        else:
            printing_copy = gs.best_params_

        print(printing_copy)

        if self.verbose:
            print('Here are all the hyperparameters that were tried:')
            raw_scores = gs.grid_scores_
            sorted_scores = sorted(raw_scores, key=lambda x: x[1], reverse=True)
            for score in sorted_scores:
                for k, v in score[0].items():
                    if k == 'model':
                        score[0][k] = utils_models.get_name_from_model(v)
                print(score)