支持向量机在电信客户流失预测中的应用研究

发布时间：2018-02-23 17:27

本文关键词： 支持向量机主成分分析客户流失预测　出处：《计算机仿真》2011年04期 　论文类型：期刊论文

【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。
[Abstract]:Customer churn analysis and prediction is an important part of customer relationship management. Due to the characteristics of telecom customers are highly nonlinear, serious redundancy and high dimension, traditional methods can not eliminate data redundancy and capture the nonlinear law, resulting in a lower prediction accuracy. In order to improve the prediction accuracy of telecom customer churn, proposes a method based on principal component analysis (PCA) support vector machine (SVM) method to predict the loss of telecom customers (PCA-SVM). The first principal component analysis of the original data, feature reduction, elimination of redundant, principal component and then obtained as a non linear input of support vector machine learning modeling. Simulation of a telecommunications company's customer data, experiment the results show that the PCA-SVM obtained the hit rate, coverage rate, accuracy rate and lift coefficient is much higher than other prediction methods. Principal component analysis with support vector machine data mining The method has a good prediction effect and provides a new method for the prediction of telecom customer loss.

【作者单位】：黔南民族师范学院计算机科学系;
【分类号】：TP18

【参考文献】