Research On Second-hand Car Valuation Model Based On General Regression Neural Network

The volume of second-hand car transactions is growing with the vigorous development of second-hand car market.China’s second-hand car trading market mainly uses the traditional second-hand car evaluation methods such as the replacement cost method.The traditional second-hand car evaluation methods have many disadvantages,such as the calculated results are not objective enough and can not be recognized and accepted by both sides of the transaction.This enlightens us to find a more accurate and objective evaluation method to be applied in the second-hand car trading market and promote the development of second-hand car companies The industry is developing healthily.Firstly,this paper summarizes the differences of used car evaluation methods used in the second-hand car market at home and abroad.According to the excellent performance of artificial neural network in data processing and prediction in various professional fields,this paper puts forward and defines the research significance of applying generalized regression neural network model to second-hand car evaluation.Secondly,eight factors are selected as the model indexes based on the hedonic price theory to study the second-hand car value evaluation.After comparing and analyzing the calculation process and prediction process of various neural networks,the new method of establishing generalized regression neural network model for second-hand car value evaluation is selected.Finally,MATLAB software is used to process the second-hand car transaction data collected in the second-hand car market,and the generalized regression neural network model and BP network model are established according to the sample data.In the second-hand car market,this paper selects the actual cases,evaluates them with generalized regression neural network model,BP network model,traditional replacement cost method and traditional current market price method,and compares the evaluation results,which verifies that the generalized regression neural network model has higher feasibility and accuracy in the application of second-hand car value evaluation.