计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (12): 26-28.

• 博士论坛 • 上一篇    下一篇

操作风险等级预测的朴素贝叶斯方法研究

王双成1,2,冷翠平1,侯彩虹1   

  1. 1.上海立信会计学院 信息科学系,上海 201620
    2.上海立信会计学院 中国立信风险管理研究院,上海 201620
  • 收稿日期:2007-12-05 修回日期:2008-01-24 出版日期:2008-04-21 发布日期:2008-04-21
  • 通讯作者: 王双成

Naive Bayes method in operational risk level prediction

WANG Shuang-cheng1,2,LENG Cui-ping1,HOU Cai-hong1   

  1. 1.Department of Information Science,Shanghai Lixin University of Commerce,Shanghai 201620,China
    2.Risk Management Research Institute,Shanghai Lixin University of Commerce,Shanghai 201620,China
  • Received:2007-12-05 Revised:2008-01-24 Online:2008-04-21 Published:2008-04-21
  • Contact: WANG Shuang-cheng

摘要: 操作风险数据积累比较困难,而且往往不完整,朴素贝叶斯分类器是目前进行小样本分类最优秀的分类器之一,适合于操作风险等级预测。在对具有完整数据朴素贝叶斯分类器学习和分类的基础上,提出了基于星形结构和Gibbs sampling的具有丢失数据朴素贝叶斯分类器学习方法,能够避免目前常用的处理丢失数据方法所带来的局部最优、信息丢失和冗余等方面的问题。

关键词: 操作风险, 等级预测, 朴素贝叶斯分类器, 丢失数据, Gibbs抽样

Abstract: It is difficult to accumulate a large number of data with high quality in operational risk.Naive Bayes classifier is the one of best classifiers used to small data set classification.It is suitable for operational risk level prediction.In this paper,firstly,the process of learning and classing is presented on naive Bayes classifier with complete data sets.Then,a method naive Bayes classifier learning with missing data is developed based on star structure and Gibbs sampling.The existing problems can be avoided in local optimization,information losing and redundancy.

Key words: operational risk, level prediction, naive Bayes classifier, missing data, Gibbs sampling