Research On Automobile Perceived Value Evaluation Based On Structural Equation Model And Neural Network

Perceived value refers to consumers’ subjective cognition of the value of products provided by enterprises based on the trade-off between income and cost,which can better reflect the degree of satisfaction of consumers’ needs.With the development of market economy and the progress of Internet technology,the types and quantity of commodities are increasing day by day,the information circulation is speeding up,consumers have more choices and the right to know,and the consumption demand is diversified.More and more enterprises realize the importance of perceived value,and evaluate the perceived value of products to find out whether they meet the needs of consumers.However,the current automobile enterprises are still constrained by the traditional marketing concept of "product centered",lack of effective evaluation of automobile perceived value and failure to timely capture the changes of market demand,resulting in a large number of models invested heavily in R &D,and many automobile enterprises are facing bankruptcy and reorganization because they can’t make ends meet.Perceived value reflects the change of consumers’ psychological state when using products,and its influence path on consumers’ acceptance(attitude and intention)is more complex and nonlinear.Traditional methods of perceived value assessment assume that the variables are linear,and can not accurately describe the causal structure among multiple variables.Structural equation model(SEM)is a widely used method of perceived value evaluation in recent years.It can verify the influence path between multiple independent variables and dependent variables,but it can not deal with the nonlinear influence between variables.Neural networks(NNs)is a mathematical analysis method that imitates human brain.It can effectively express the nonlinear relationship between variables,but there is no explanation for the structural path between neurons.Therefore,using the path definition ability of SEM and the nonlinear mapping ability of NNs,this paper innovatively proposes a modeling method combining SEM and NNs to improve the accuracy of automobile perceived value evaluation,help automobile enterprises better identify consumer demand and improve product competitiveness.The main research contents are as follows:This paper constructs a driving model of automobile perceived value,which is composed of four dimensions and 15 driving factors.Combining with the durable goods perceived value model and automobile consumption trend,the driving factors of automobile perceived value are deeply mined.Secondly,the structure of the evaluation model of automobile perceived value is defined: the path relationship between automobile perceived value and consumer acceptance is determined by SEM,and the main driving factors affecting perceived value are screened out.Thirdly,the improvement of vehicle perceived value model is completed: the path analysis results of SEM are transformed into the topology structure of NNs,and the BP(back propagation)algorithm of NNs is used to train the sem-nns model to improve the convergence validity and prediction accuracy of the model,and obtain the connection weight value between the neuron nodes.Fourthly,the superiority of sem-nns modeling method is verified.By comparing and analyzing the goodness of fit of SEM model and sem-nns model and the path effect among model variables,it is proved that the goodness of fit and path effect of senm NNs model are more interpretable.Fifthly,it shows the application case of the car perceived value evaluation model: it evaluates the perceived value of specific brand models,and verifies the output of the model combined with the actual market sales.