libscores.py 文件源码

def compute_all_scores(solution, prediction):
    ''' Compute all the scores and return them as a dist'''
    missing_score = -0.999999
    scoring = {'BAC (multilabel)':nbac_binary_score, 
               'BAC (multiclass)':nbac_multiclass_score, 
               'F1  (multilabel)':f1_binary_score, 
               'F1  (multiclass)':f1_multiclass_score, 
               'Regression ABS  ':a_metric, 
               'Regression R2   ':r2_metric, 
               'AUC (multilabel)':auc_metric, 
               'PAC (multilabel)':npac_binary_score, 
               'PAC (multiclass)':npac_multiclass_score}
    # Normalize/sanitize inputs
    [csolution, cprediction] = normalize_array (solution, prediction)
    solution = sanitize_array (solution); prediction = sanitize_array (prediction)
    # Compute all scores
    score_names = sorted(scoring.keys())
    scores = {}   
    for key in score_names:
        scoring_func = scoring[key] 
        try:
            if key=='Regression R2   ' or key=='Regression ABS  ':
                scores[key] = scoring_func(solution, prediction)
            else:
                scores[key] = scoring_func(csolution, cprediction)
        except:
            scores[key] = missing_score
    return scores