一种基于粗集的欺诈风险分析方法

Diagnosis on the Application of Rough Set Theory in Cheat Risks

文章利用粗集理论中的特征属性约简方法来分析和研究欺诈风险 ,提出了一种基于粗集的欺诈风险分析方法 ,并通过一个信用卡欺诈的分析实例对该算法进行了检验。实验结果表明 ,在保证分类质量基本不变的情况下 ,该算法是有效的 ,它能找出引发欺诈行为的主要特征属性 ,帮助管理者及时、准确地预测各种可能发生的欺诈风险。

In order to compress or reduce redundant features in cheat risks diagnosis,the rough set theory is introduced and a feature reduction algorithm based on the rough set is proposed. An example in cheat risks diagnosis is given to validate the algorithm. The results show that when the cheat classification result is almost invariable,the main features which are more important to the cheat classification can be searched by this algorithm.