Measuring Customer Satisfaction based on Neural Networks Partial Least Squares Approach

Customer satisfaction measurement is an important part of marketing research in industrial organizations since it is the key to formulating customer value strategies and to continuously improving implementation of these strategies. Traditional techniques for modeling the network such as partial least squares (PLS) lack the capability of fitting the nonlinear and asymmetric relationships. This article presents a new technique of neural networks partial least squares (NNPLS) to measure customer satisfaction. The details of NNPLS are discussed. The results show that the NNPLS gives the smaller prediction errors compared with linear PLS. Therefore a robust model expressed by NNPLS succeeds in correlating the relations between customer satisfaction, customer loyalty and their drivers