历史数据缺失情况下银行信贷决策研究——基于谨慎信任场和机器学习的模型构建

针对传统信贷业务决策模型难以评估新增客户状况、研究角度单一等问题,本文结合信誉和基于信任的合作机制,构建了一个融合谨慎阻力、机器学习等理论的银行信贷业务决策模型.具体做法如下:首先,通过谨慎信任场模型计算得到信任密度和阈值对比,据此淘汰部分信贷客户;然后,用机器学习预测问题客户,进行二次筛选;最后,比较精选客户的信任压强,进而为信贷业务决策提供参考.实验结果证实,谨慎信任场模型能够有效剔除问题企业,对于指导银行信贷业务决策具有重要意义.

Bank Credit Decisions in the Absence of Historical Data:Based on Prudent Trust Field and Machine Learning

In view of the traditional credit decision-making models difficult to evaluate the situation of new customers and the incomprehensive research perspective, a bank credit decision-making model integrating trust cooperation mechanism, prudent trust field, machine learning and other theories has been constructed. The specific methods are as follows: firstly, calculate the trust density by prudent trust field model and compare the trust density with the threshold value to eliminate some credit customers;secondly, machine learning is used to predict and eliminate the problem customers among these remained customers;finally, compare the trust pressure among these selected customers to provide a reference for credit decisions. The experimental results show that the prudent trust field model can effectively eliminate problematic enterprises, and it would have great significance to guide the decision of bank credit.