期刊文献+

基于数据挖掘技术的银行客户定期存款认购模型研究

A Research on Model of Statistical Decision on Research for Bank’s Long-term Deposit Subscription Based on Data mining Algorithm
收藏 分享 导出
摘要 数据挖掘技术能有效地挖掘出潜在的银行客户,能够大大提高银行的竞争力.介绍了数据挖掘技术中常用的三种模型:逻辑回归模型、BP神经网络模型和决策树模型,同时构造了一种新模型--逻辑回归与BP神经网络混合的模型,然后分别采用这四种模型对可能影响银行客户是否认购定期存款的影响因素进行数据挖掘分析,分别构建了基于逻辑回归模型、BP神经网络模型、逻辑回归与BP神经网络的新模型、决策树模型的银行客户定期存款认购的四种模型,同时利用R语言分别对这四种模型进行分析,分别用ROC曲线的AUC值和正确率比较这四种模型的功效强弱以及稳定性,研究结果表明,给出的新模型--逻辑回归与BP神经网络的新模型的预测效果更好,训练集和测试集预测的准确率分别为0.936和0.931,训练集和测试集ROC曲线的AUC值分别为0.998和0.987,这可以大大缩小银行推送认购定期存款的客户范围,有效地挖掘出潜在的银行客户,可以大大提高银行的效率. Data mining technology can dig out potential bank customers and improve the competitiveness of Banks.Firstly,three common models(logistic regression,artificial neural network BP and decision tree) are introduced,moreover a new mix model of logistic regression and artificial neural network BP is constructed in this paper.Secondly,effect factors of bank’s long-term deposit subscription are analyzed by the four models,and four models of bank’s long-term deposit subscription axe constructed basing on logistic regression,artificial neural network BP,new model and decision tree by R.Thirdly,Based on the analysis of accuracy rates and AUC,the four models are compared on the stability and efficacy.At last,results show that the new model has better predicting result than other three predict methods,the accuracies of train and test are 0.936 and 0.931 respectively,the AUCs of train and test are 0.998 and 0.987 respectively,such knowledge greatly reduces the market in grange of potential clients for term deposit thus improves the bank efficiency.
作者 张利利 郭淑妹 马艳琴 卜春霞 ZHANG Li-li;GUO Shu-mei;MA Yan-qin;BU Chun-xia(Huanghe Science and Technology College,Zhengzhou 450063,China;Strategy PLA Information Engineering University,Zhengzhou 450001,China;Zhengzhou University,Zhengzhou 450001,China)
出处 《数学的实践与认识》 北大核心 2019年第21期95-102,共8页 Mathematics in Practice and Theory
基金 河南省科技厅基础与前沿技术项目(182102311100)。
关键词 逻辑回归 BP神经网络 决策树 新模型 正确率 ROC曲线 logistic regression artificial neural network BP decision tree new model accuracy rate ROC curve
  • 相关文献
投稿分析

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部 意见反馈