基于SVM分类器的电信欠费预测模型

对于日益严重的电信欠费问题,采集了某个地区电信的呼叫详单记录、客户信息、欠费和交费信息等数据,使用支持向量机建立了欠费预测模型.然后可以利用所建立的模型来预测潜在的欠费客户,决策者可以得到充分的支持,做出正确的决策.为了提高模型的预测准确率,使用了双变量统计和主成份对数据进行预处理和分析.最后,为了得到最好的模型,主要做了三种实验.实验结果表明使用SVM建立的模型具有很好的预测准确率.

An arrear predicting model based on support vector machine in telecom

CDR(call detail record), customer information, paying and arrear information in telecom in some area are collected to solve the progressive arrear problem. SVM(Support Vector Machine)was used to construct an arrear predicting model in telecom. Then the model can be applied to predict potential arrear customers in telecom and the decision-maker can receive efficient support and make right decision. To improve the forecasting accuracy, bivariate statistics and PCA(principal component analysis) were used to prepare and analyze the data in advance. At last, to get the best model, three kinds of experiments were done. The results show that it has fine predicting results with SVM model.