难点一：如何在客观赋权时考虑权重的大小反映中小型企业客户违约状态的差异程度。现有研究通过灰色关联度、均方差、变异系数等方法对指标进行赋权这种方式或是反映与均值的差异程度、或是反映与理想值的差异程度或是反映任意两个评价对象之间的差异程度但是没有考虑权重的大小是否能够显著区分违约客户和非违约客户的差异程度。难点二：如何在组合赋权时考虑组合权重的大小反映中小型企业客户违约状态的差异程度。现有研究通过极差

难点一：在客观赋权时考虑权重的大小反映中小型企业客户违约状态的差异程度的问题尚未得到很好的解决。现有研究使用灰色关联度、均方差、变异系数等方法对指标进行赋权，但这些方法只能反映指标与均值、理想值或其他评价对象之间的差异程度，并未考虑权重的大小是否能够显著区分违约客户和非违约客户的差异程度。

难点二：在组合赋权时考虑组合权重的大小反映中小型企业客户违约状态的差异程度的问题也尚未得到很好的解决。现有研究通过极差最大、信用得分差异最小等方法对指标进行组合赋权，但这些方法只能反映指标与均值、理想值或其他评价对象之间的差异程度，并未考虑不同类型客户之间的差异程度。

In academic terms:

Challenge 1: There is a need to consider the significance of weights in objectively assigning values that reflect the degree of default status differentiation among small and medium-sized enterprise (SME) customers. Existing research employs methods such as grey correlation, mean square deviation, and coefficient of variation to assign weights to indicators. However, these methods only reflect the differences between indicators and the mean, ideal value, or other evaluation objects, without considering whether the weights can significantly differentiate between default and non-default customers.

Challenge 2: There is a need to consider the significance of weights in combining indicators to reflect the degree of default status differentiation among SME customers. Existing research utilizes methods such as maximum range and minimum difference in credit scores to combine weights for indicators. However, these methods only reflect the differences between indicators and the mean, ideal value, or other evaluation objects, without considering the differentiation among different types of customers