基于异常检测和神经网络的财政欺诈屏蔽分析

Analysis of the Financial Fraud Screening Based on the Anomaly Detection and Neural Network

随着国家对"三农"问题的重视,作为"三农"政策的重要组成部分,近年来国家逐渐加大农业财政补贴的力度,同时也出现一些财政补贴申请存在欺诈的问题。以Clementine提供的虚拟数据为基础,分析财政补贴申请中可能出现欺诈行为的情况,通过运用SPSS Clementine 11.1软件,利用异常检测和神经网络两种分类算法,对财政申请的欺诈行为进行数据挖掘分析,挖掘出存在较大欺诈可能性的申请者。

As the country＇s emphasis on ＂three rural＂ issue, as an important part of the ＂three rural＂ policy, in recent years, the government gradual- ly increases the intensity of agricultural subsidies. At the same time there are also appeared some subsidies fraud. Based on the virtual data provided by Clementine, analyses the fiscal subsidy application may occur in the case of fraud, by using the software of SPSS Clementine 11.1, uses two kinds of classification algorithms include anomaly detection and neural network, carries on the data mining analysis to the financial application fraudulent practice, digs out the possibility of applicants is fraud, finally digs out the applicants which with big possibility of fraudulent.