Shopping Website Assessment Based On BP Neural Network

With the popularity of the Internet, online shopping for its convenience plays a more and more important role in people’s life, and the number of online shopping is increasing year by year. Due to the increasing demand rate, shopping sites available for public consumption greatly increased in number and variety, but the quality and level of the shopping website is good and bad are intermingled. Now about shopping website evaluation theory knowledge is gradually perfect, for shopping website also more and more empirical cases. Online shopping is not only to the operation of the site experience, more important is behind the website services provided by the experience. Therefore evaluate the shopping website work not only has a certain theoretical significance, but also has great practical significance.In this article, through learning website evaluation index system and evaluation methods both at home and abroad research results, with a certain theoretical basis to choose the level analysis and neural network methods, established is suitable for the technical route, this article in constructing evaluation index system of good after finished the design of the evaluation model, and uses this model to conduct empirical research and result analysis. Main achievements include the following three aspects:first, using literature research, expert consultation method to establish the initial indicator, through the questionnaire to obtain data after factor analysis, select the appropriate indicators and carries on the classification, and then the weights are calculated by analytic hierarchy process (ahp), to complete the shopping website construction of the evaluation index system. Second, three layer BP neural network theory in control method and operation method, on the basis of MATLAB to complete the network structure of the model design and program design of the input layer contains 20 neurons,12 neurons in hidden layer contains, the output layer contains a neuron, and determine the parameters for the network training by repeated experiments, complete the construction of network evaluation model. Through the expert scoring method and software testing method in this paper, the selected 30 shopping site specific rating,20 web site before use as the training sample, after 10 sites as test samples, the model of learning and training simulation test and calculate the simulation error, prove accurate evaluation model. Third, Choose five infants and pregnant shopping sites for empirical research, introduced in detail the evaluation index system and evaluation model using method, the specific steps and to evaluate the ranking analysis of fully validated the practical significance of this study, and finally puts forward suggestions for improvement to the shopping website operators and managers, investors and entrepreneurs respectively.