Title :
CostBP Algorithm and its Application in Customer Churn Prediction
Author :
Dong, Yan-Jie ; Wang, Xue-Hua ; Zhou, Jun
Author_Institution :
Sch. of Manage., Dalian Univ. of Technol., Dalian, China
Abstract :
For data sets of customer churn are imbalanced, it´s hard for classical BP neural network to make good classification. This paper proposes an improved BP algorithm, namely CostBP. CostBP modifies the error function into cost-based error function, based on which error propagation is conducted. The experimental results showed that CostBP can minimize total misclassification cost and was effective in prediction of minority class.
Keywords :
backpropagation; customer relationship management; BP neural network; CostBP algorithm; cost-based error function; customer churn prediction; Boosting; Classification algorithms; Conference management; Costs; Customer relationship management; Neural networks; Predictive models; Support vector machine classification; Support vector machines; Technology management; CostBP; churn prediction; imbalanced data sets;
Conference_Titel :
INC, IMS and IDC, 2009. NCM '09. Fifth International Joint Conference on
Conference_Location :
Seoul
Print_ISBN :
978-1-4244-5209-5
Electronic_ISBN :
978-0-7695-3769-6
DOI :
10.1109/NCM.2009.222
