银行信用风险预测——基于SVM和BP神经网络的比较研究  

Bank Credit Risk Forecast——Based on SVM and BP Neural Network

作　　者：李佳[1] 黄之豪 LI Jia;HUANG Zhihao

机构地区：[1]上海理工大学,上海200093

出　　处：《上海立信会计金融学院学报》2018年第6期40-48,共9页Journal of Shanghai Lixin University of Accounting and Finance

基　　金：上海理工大学人文社科"攀登计划"项目(SK18PB04)

摘　　要：本文使用支持向量机和BP神经网络两种人工智能方法对2015-2016年间我国144家沪深两市上市公司进行研究,考察人工智能方法在商业银行构建信用风险预测模型中的作用。研究结果表明:SVM模型和BP神经网络模型都具有良好的预测能力,SVM模型在性能和预测精度上略优于BP神经网络模型。In this paper,144 listed companies in Shanghai and Shenzhen between 2015 and 2016 are analyzed with support vector machines and BP neural network to explore the role of artificial intelligence in construction of credit risk prediction models for commercial banks.It shows that:Both SVM model and BP neural network model have good prediction ability.The SVM model is slightly better than BP neural network model in performance and prediction accuracy.

关 键 词：人工智能 支持向量机 BP神经网络 信用风险 

分 类 号：F832.4[经济管理—金融学]