摘要
对于客户获取策略中客户反应行为模式分析,从数据挖掘技术的角度可以归结为分类问题·依据组合分类方法的思想,提出一种基于遗传算法的多重决策树组合分类方法来提高分类的准确性和精确度·该组合分类方法将以概率度量水平的多重决策树进行并行组合,采用遗传算法优化连接权值矩阵·在仿真分析中采用二元反应行为模式的客户反馈仿真数据对该组合分类方法进行测试和评估·实验结果表明,在保持分类结果良好可解释性的基础上相比于单个决策树方法,该组合分类方法具有更高的分类精度,并优化了分类规则·
The analysis of customer response behavior patterns in customer winning strategy can be attributed to a problem of classification from the view point of data mining. Keeping the combination classification procedure in view, a multiple decision tree approach to higher exactness and accuracy of classification is set out on a genetic algorithm basis. The approach will form combination in parallel though the multiple decision tree in accordance to the level of probability measure and use genetic algorithm to optimize the connected weight matrix. customer feedback data are taken as binaryresponse behavior patterns in simulating analysis to test and evaluate the new approach. The testing results showed that the accuracy of such a classification approach is higher than that of single decision tree approach, with the classifying rules optimized and the explainability kept for classification results.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第11期1112-1115,共4页
Journal of Northeastern University(Natural Science)
基金
辽宁省自然科学基金资助项目(9910200208).
关键词
遗传算法
多重决策树
组合分类方法
客户反应行为模式
数据挖掘
multiple decision tree
genetic algorithm
combination classification approach
customer response behavior patterns
data mining
