基于判别分析法和多层感知器神经网络的大数据评价模型研究  

Research on Big Data Evaluation Model Based on Discriminant Analysis and Multilayer Perceptron Neural Network

作　　者：田青青[1] 吴惠珍[1] TIAN Qing-qing;WU Hui-zhen(Medical College of Chuzhou City Vocational College,Chuzhou,Anhui 239000)

机构地区：[1]滁州城市职业学院,安徽滁州239000

出　　处：《怀化学院学报》2022年第5期42-47,共6页Journal of Huaihua University

摘　　要：通过大数据提取影响信用状况的各种信息因素包括个人的基本信息、已发生的借贷和偿还、信用透支额度等方面进行分析,建立基于大数据的信用评价模型.利用判别分析法和多层感知器神经网络分析法分别对个人信用建立模型并进行比较评价.首先将得到的数据做清洗工作,剔除与信用评价影响不大的指标变量,再引入与信用评价有关的指标,在大数据基础上建立了2个信用评价模型,最后利用SPSS(Statistical Product and Service Solutions,SPSS)软件将数据代入模型,得到信用评价的结果.结果表明,多层感知器的分类结果优于判别分析的分类结果.Through the extraction of various information factors that affect credit status through big data,the basic information of individuals, the loan and repayment that have occurred, and the amount of credit overdraft can be analyzed,and a credit evaluation model based on big data can be established. Using discriminant analysis and multi-layer perceptron neural network analysis, the personal credit models are established, compared and evaluated. Firstly, the obtained data is cleaned, the index variables that have little influence on credit evaluation are eliminated, and the indicators related to credit evaluation are introduced. Secondly,two credit evaluation models are established on the basis of big data. Finally,by means of the SPSS(statistical product and Service solutions) software,the data is integrated into the model to get the result of the credit evaluation. The results show that the classification results of the multilayer perceptron are better than the classification results of the discriminant analysis.

关 键 词：大数据 信用评价模型 判别分析 神经网络 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]