用LDA Boosting算法进行客户流失预测  

Research on the LDA Boosting in Customer Churn Prediction

作　　者：应维云[1] 蔺楠[1] 谢雅雅[2] 李秀[2] 

机构地区：[1]上海财经大学,上海200433 [2]清华大学国家CIMS工程研究中心,北京100084

出　　处：《数理统计与管理》2010年第3期400-408,共9页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(70671059)

摘　　要：本文提出一种LDA boost(Linear Discriminant Analysis boost)分类方法,该算法能有效利用样本的所有特征,并且能够从高维特征空间里提取并组合优化出最具有判别能力的低维特征,使得样本类间离散度和类内离散度的比值最大,从而不会产生过度学习,大大提高算法效率。该算法有效性在某商业银行的客户流失预测过程的真实数据集中得到了验证。与其他同类算法,如人工神经网络、决策树、支持向量机等运算结果相比,该方法可以显著提高运算精度。同时,LDAboosting与其他boosting算法相比,也具有显著的优越性。In this paper,a novel classification algorithm called LDA boosting is proposed to predict the customer churn.This algorithm can effectively extract and assemble the most discriminative low-dimensional features from all the features in the high-dimensional feature space of the samples,which make the maximal ratio of intra-class dispersion and inter-class dispersion,thus will not produce exhaustive search,and will greatly improve the learning efficiency.The effectiveness of the proposed algorithm is validated by churn prediction experiments on a real bank customer churn data set.The method is found to improve prediction accuracy significantly compared with other algorithms,such as artificial neural networks,decision trees,and support vector machines.Moreover,LDA boosting also produces better prediction results than other boosting algorithms.

关 键 词：客户流失 数据挖掘 预测 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] F270[自动化与计算机技术—控制科学与工程]