Research On The Classification Of Customer’s Value Based On The RBF Neural Network

With the development of economy and society, competition between ports is becoming increasingly fierce, if the port enterprise wants to survive and develop, they must set up the management idea as’take the customer as the center’. However, "take the customer as the center" is not taking all customers as the center, enterprise should regard the customer who bring the greatest value to them as the center. Therefore, it is particularly important for enterprises to enhance their corporate profitability and core competitive ability according to analyze the customer value and carry on the classified management on the basis of different customer value. As there is a nonlinear relationship between the customer classification indexs, the Radial Basis Function (RBF) neural network shows good performance in dealing with nonlinear problem. Therefore, this paper uses the RBF neural network to classify customers.This paper proposes a research on the classification of customer’s value based on the RBF neural network and regard the port customer as the research object. This paper mainly done the following research:Firstly, under the guidance of customer value theories, this paper analyze the process of port business and build the index system of port customer value evaluation; secondly, through analyze the advantages and disadvantages of different customer classification methods, and considering the nonlinear relationship between customer indexs, this paper selected RBF neural network as the port customer classification method; thirdly, this paper extract the customer index data from the database of port production business system and regard them as the input data of RBF neural network model, at the same time, using entropy value method to determine indexs weight and regard the composite score as the expected output of the RBF neural network model; fourthly, through optimize the RBF neural network algorithm, this paper construct the RBF neural network model,and then simulate and validate the model; finally, inputting the other unclassified customer index data into the constructed model, analyzing its error to further verify the effectiveness of the network model, at the same time, analyzing the customer classification results and index data to summarize the port customer characteristics and the cause of the problem and put forward the improvement measures and suggestions.