电子商务客户流失的建模与预测研究

Research on E-business Customer Churning Modeling and Prediction

研究电子商务客户流失预测问题,电子商务客户流失具有非线性、时变等特点,用单一预测模型难以对电子商务客户流失变化规律进行全面、准确预测,导致预测正确率低。为了提高电子商务客户流失预测正确率,提出一种组合的电子商务客户流失预测模型。组合预测模型首先采用遗传算法对影响客户流失因子进行筛选,提取对预测结果影响重要的因子,然后分别采用支持向量机和神经网络对其进行预测,最后采用支持向量机对两种预测结果进行融合,得到组合模型的电子商务客户流失预测结果。仿真结果表明,组合模型提高了电子商务客户流失预测正确率,解决了单一预测模型的缺陷,将为电子商务客户流失研究提供一种新预测思路。

Study electronic commerce customer churning prediction problem.The electronic commerce customer data change is non-linear and time-varying and other characteristics,using a single prediction model to accurately predict e-commerce customer loss is difficult.In order to improve the prediction accuracy rate of electronic commerce customer churning,the paper put forward a kind of combination forecasting model of electronic commerce customer churning.The model first used the genetic algorithm for the screening of effecting factors,and extracted the important influence factors which effect the predicting results.Then support vector machine and neural network were respectively used to carry out the forecast.Finally,by using support vector machine to fuse the two prediction results to acquire the prediction results of the combination model.Simulation results show that the combined model can improve the prediction accuracy rate of the electronic commerce customer churning,and provides a new prediction method for the electronic commerce customer churning.