支持向量机在电信客户流失预测中的应用研究  

Study on Telecom Customer Leaving Prediction Based on Support Vector Machine

作　　者：王观玉[1] 郭勇[1] 

机构地区：[1]黔南民族师范学院计算机科学系,贵州都匀558001

出　　处：《计算机仿真》2011年第4期115-118,312,共5页Computer Simulation

摘　　要：客户流失分析与预测是客户关系管理的重要内容。由于电信客户的特征呈高度非线性、严重冗余和高维数,传统方法无法消除数据之间冗余和捕获非线性规律,导致预测精度较低。为了提高电信客户流失预测精度,提出一种基于主成份分析(PCA)支持向量机(SVM)的电信客户流失预测方法(PCA-SVM)。首先利用主成分分析对原始数据进行特征降维,消除冗余,然后将得到的主成分作为非线性支持向量机的输入进行学习建模。对某电信公司客户流失数据进行了仿真,实验结果表明,PCA-SVM获得的命中率、覆盖率、准确率和提升系数远远高于其它预测方法。说明主成分分析结合支持向量机的数据挖掘方法具有很好的预测效果,为电信客户流失预测提供了一种新方法。Customer Leaving analysis and prediction is an important content of the customer relationship management.Features of Telecom Customer data are highly redundant and nolinear,therefore,traditional method cannot eliminate data redundancy and draw the nonlinear rule,and the prediction accuracy is very low.In order to improve the accuracy of telecom customer leaving prediction,a new method is proposed based on principal component analysis（PCA） and support vector machine（SVM） in this paper.The original high dimensional is lowered by principal component analysis and principal components are determined.The low dimensional data sets are used as the inputs of support vector machine predictor.The experimental results of customer leaving prediction for a telecommunication carrier show that the PCA-SVM method is superior to traditional method in hit rate,covering rate,accuracy rate and lift coefficient.This research indicates that the data mining method of PCA-SVM has a good prediction effect,and can work as a new method for customer leaving prediction.

关 键 词：支持向量机 主成分分析 客户流失 预测 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]