结合测度整合和AWCBA算法的个人信用评估研究

Personal Credit Evaluation Combining Measure Integration and AWCBA Algorithm

在个人信用评估领域,对个人信用等级的预测是最具有挑战性的。提高对个人信用等级预测的准确度可以避免大量死账坏账的出现。然而传统的个人信用评估模型假设全部属性都具有相同的重要性,并且使用单一测度进行规则的剪枝和预测,这些个人信用评估模型往往太主观,不能取得较好的分类效果。该研究结合测度整合和(Adaptive Weighted Classification Base of Association,AWCBA)算法构建了个人信用评估模型,对客户基础属性进行自适应化加权,并引用了支持度、置信度和卡方测度的调和均值作为分类依据,实现个人信用等级的分类,与其他算法相比,AWCBA算法预测准确度比其他算法都要高。

In the field of personal credit evaluation,the prediction of personal credit rating is the most challenging.Improving the accuracy of personal credit rating prediction can avoid a large number of bad debts.However,the traditional personal credit evaluation models assume that all attributes are of the same importance,and use a single measure to prune and predict the rules.These personal credit evaluation models are often too subjective to achieve better classification results.Thus,a personal credit rating model is constructed based on the measurement integration and the Adaptive Weighted Classification Base of Association (AWCBA) algorithm.The basic attributes of customers are weighted adaptively,and the harmonic mean of support,confidence and chi-square measures are used as the classification basis to realize the classification of personal credit rating.Compared with other algorithms,the AWCBA algorithm has a higher prediction accuracy.