Using Artificial Neural Networks to Establish a Customer-cancellation Prediction Model

Back propagation neural networks and general regression neural networks are applied to establish a customer-cancellation prediction model and the empirical results showed that both prediction models possessed good predictive abilities and can aid in service capacity scheduling.

Abstract

In the past, judgments concerning customer cancellations relied primarily on managers' experience. Prediction errors can cause surpluses or insufficient service capacity. Data mining technology can improve prediction and judgment accuracy. This study applies back propagation neural networks and general regression neural networks to establish a customer-cancellation prediction model. The empirical results showed that both prediction models possessed good predictive abilities and can aid in service capacity scheduling.