基于数据挖掘的商户风险评分方法和系统

本文提供了一种商户风险评分系统和方法,可根据商户的历史交易行为有效地对商户欺诈风险进行风险评分.针对发生过一类欺诈交易的商户进行特征行为分析,提取风险指标变量;对风险指标变量进行相关性分析,剔除相关性高的冗余变量;之后根据分析结果定义正常商户与高风险商户规则;按规则重新提取高风险商户及正常商户样本;接着采用决策树模型进行训练,通过训练好的决策树模型计算所有样本变量的输出,作为一个新的风险指标变量;随后,对新的样本数据采用逻辑回归算法训练模型.使用上述的方法针对各类欺诈交易商户分别进行训练建模.最后将商户样本数据分别输入各个模型,进行加权平均,得到商户的风险评分结果.

Merchant Risk Evaluation Method and System Based On Data Mining Technology

This paper presents a merchant risk evaluation method and system based on data mining technology. We analyze merchant behavior characteristics by comparing the transaction differences between normal merchants and fraud merchants, therefor define merchant risk characteristic variables and fraud rules, and then preprocess transaction data with decision tree model, score merchant risk of each type with logistic regression model, and finally evaluate merchant total risk by calculating weighted average of merchant risk scores with different risk types.