将数据挖掘方法与金融领域知识相结合,研究可疑金融交易识别策略与方法,是我国反洗钱领域的重要基础性工作。决策树分析是一种重要的数据挖掘方法,通过对案例数据的训练学习达到对未知类标识数据的分类。基于决策树分析的思想设计出适合于可疑外汇交易识别的CART分类方法,并用真实外汇交易数据对该方法进行了验证,实验结果表明该方法可有效提高可疑金融交易识别效率。
In this paper, we have analyzed the characteristics of Chinese foreign exchange money laundering activities, and combined the decision tree approach with financial domain knowledge. We chose the suitable money laundering transaction recognition strategy and method. By making full use of the real transaction data to perform the experiment, useful rules of money laundering were discovered. Experimental results on SAS demonstrate that CART algorithm can be extremely useful in money laundering recognition.