Abstract: To reduce the loss of innovative customers churn for collaborative product innovation process and enterprise,and to identify the innovative customers intended to leave in collaborative products innovation effectively,the improved cost sensitive support vector machines model was put forward to predict the innovative customer churn.On the basis of analyzing the characteristics of the innovative customer churn prediction,the research framework of innovative customer churn prediction was presented.Aiming at the problem that most of the traditional classification algorithms taking the predict accuracy as goal,which ignored the cost of customer churn classification as errors and the unbalance of dataset,the innovative customer churn prediction model based on improved cost sensitive support vector machines was built.As indicated by the results of application,the validity and feasibility of proposed prediction model was indicated.

Key words: collaborative products innovation, innovative customer, churn prediction, support vector machines

摘要： 为了降低创新客户流失对协同产品创新过程及企业造成的损失,在协同产品创新中有效识别具有流失倾向的创新客户,提出改进的代价敏感学习支持向量机创新客户流失预测模型。在分析创新客户流失预测特点的基础上,提出创新客户流失预测研究框架;针对传统分类算法大多以预测精度为目标、忽略创新客户流失误分类的代价和数据集非平衡性的问题,建立了基于改进代价敏感支持向量机的创新客户流失预测模型。通过应用结果表明了该预测模型的有效性和可行性。

关键词: 协同产品创新, 创新客户, 流失预测, 支持向量机