大数据背景下基于GRBF神经网络的P2P机构信用风险预警研究  

Research on Credit Risk Early Warning of P2P Institutions Based on GRBF Neural Network in The Background of Big Data

作　　者：李桂芝 王雪标[1] LI Gui-zhi;WANG Xue-biao(Dongbei University of Finance And Economics,Dalian 116000,China;Yingkou Institute of Technology,Yingkou 115000,China)

机构地区：[1]东北财经大学经济学院,辽宁大连116000 [2]营口理工学院经济管理学院,辽宁营口115000

出　　处：《数学的实践与认识》2020年第17期35-43,共9页Mathematics in Practice and Theory

基　　金：营口理工学院创新团队支持计划(IRTYKIOT)。

摘　　要：针对P2P机构信用风险预警问题,提出了基于大数据思维的信用评估体系,采用基于动态特征的广义径向基神经网络对228家P2P机构12个月的高维数据指标进行信用风险评估.应用设计的广义径向基神经网络和BP神经网络进行对比,准确率分别为91.9%、85.2%,广义径向基神经网络在处理实时高维数据时表现出良好的性能,可以对我国P2P机构信用风险进行预警.同时深入对预警机构进行数据分析发现,如果企业资金流动性较差、净流入低也可能存在较高风险,企业应依据小额分散的借贷原则,降低借款集中度可以有效防范企业信用风险.Aiming at the credit risk early warning of P2P institutions,a credit evaluation system based on big data thinking is proposed.In this paper,the generalized radial basis function(GRBF)neural network based on dynamic characteristics is used to evaluate the credit risk of 228 P2P institutions with 12-month high-dimensional data.The results show that the improved GRBF neural network has better predicting result than BP neural network,the accuracies of GRBF and BP?are 0.919 and 0.852 respectively,the GRBF neural network has good performance in dealing with real-time and high-dimensional data,and it can early warn the credit risk of P2P institutions in China.At the same time,the data mining of credit early warning enterprises shows that if the liquidity of the enterprise is poor and the net inflow is low,there may be higher risk,According to the principle of small and scattered loan,enterprises should reduce the loan concentration to effectively prevent the credit risk of enterprises.institutions in China.At the same time,the data mining of credit early warning enterprises shows that if the liquidity of the enterprise is poor and the net inflow is low,there may be higher risk,According to the principle of small and scattered loan,enterprises should reduce the loan concentration to effectively prevent the credit risk of enterprises.

关 键 词：广义径向基神经网络 P2P机构 信用风险 

分 类 号：F724.6[经济管理—产业经济] F832.4[自动化与计算机技术—控制理论与控制工程] TP183[自动化与计算机技术—控制科学与工程]