Neural Network with Forgetting: An ANN Algorithm for Customer

Abstract

As a useful analytical tool, artificial neural networks (ANN) are widely applied in analyzing the information stored in enterprise database or data warehouse. Certain structured ANN can also be used for customer segmentation. In the process of establishing the training data sets, there is the problem of selecting customer data from potentially wide-varying time periods, as is in a typical enterprise data warehouse. To solve this problem, we introduced a forgetting coefficient into the BP ANN; we also established a new ANN, ANN with forgetting, to improve the customer segmentation process.