Previous studies often binarized credit ratings, but their impact on machine learning model performance has not been thoroughly examined. [6] addressed this by evaluating different binarization methods with SAFE metrics [1]. In this paper, we need to analyze various classifications and identify BB = 1 and BBB = 1 as the best two classifications through entropy, balance scores and transition matrices. Confusion matrix, ROC curve and predicted probability distribution are then used to identify the best-performing classification scheme relative to the alternatives. This study validates the innovativeness and applicability of SAFE metrics, highlighting their effectiveness in distinguishing between high-risk and low-risk classifications.