基于属性约简的决策树算法在商品促销决策中的应用研究

商品销售管理是企业经营管理中的一个重要环节，通过建立商品促销决策的数据挖掘模型，企业可以获取有助于销售决策的大量有价值的信息，从而获得竞争优势.该文采用决策支持模型来发掘不同客户群的商品购买需求.首先对收集的历史数据进行预处理，然后采用似然比计算预测字段和目标字段的相关性，对属性进行约简，最后使用CHAID算法构建决策模型.实验结果表明，约简模型的性能降低很小，但是大大提高了运行效率和分析效率，降低了过度拟合模型的风险，具有很好的实用价值.

Application Research in Sales Promotion Decision Based on Decision Tree Algorithm of Attribute Reduction

Merchandising management is an important part of enterprise operation and management. Through establishing data mining models of sale promotion decision, enterprises catch a lot of valuable and helpful information that help the sales in decision-making, and accordingly build competitive advantage. This paper presents decision-making model to uncover the unique information needs of users. Firstly, collected historical data from customers is preprocessed, and then attribute dimension reduction is implemented according to independent testing of the prediction field and target field, finally, decision rules are generated with CHAID algorithm. Experimental results show that reduction model decreases little performance, but improves efficiency of running and analysis, reduces the risk of over-fitting model and has good practical value compared with no reduction model.