The Value Evaluation Research Of Used Car Based On Random Forest Model

With the development of Chinese economy,the automobile market is booming day by day.At the same time,more and more consumers accept the concept of used car.By the end of 2019,Chinese used car trading volume has increased a lot,reaching 14.9228 million,which not only promotes the development of used car industry,but also puts forward higher requirements for the value evaluation of used car.However,due to the unbalanced development of the used car market in different regions of China,the use of the three traditional evaluation methods will be limited by the lack of evaluation data,the asymmetry of evaluation information and the high cost of evaluation process.Therefore,it is very important to find an evaluation method that can meet the needs of Chinese used car market.As a machine learning model,random forest,which has unique advantages in data processing such as being able to cope with noise and data loss in the sample,is realized by computer algorithm.Therefore,this paper introduces random forest model in the used car value evaluation research.First,according to the importance order which realized by the random forest model,ten feature variables are selected to establish the characteristic variable system of used car value evaluation.These feature variables can be divided into three categories,entity variable,function variable and market variable according to different attributes.Second,the samples are divided into training set and test set according to the proportion of 8:2.And there are two important parameters in the random forest model,mtry and ntree,which will affect the final results.The used car value evaluation model based on random forest is established according to the training set data and the characteristic variable system after the two optimal parameters are determined through multiple tests.Third,using test set to check the evaluation accuracy of the model by index measurement and error analysis.There are four indexes,goodness of fit,mean absolute error,mean relative error and root mean square error are selected to grasp the overall operation of the model.Through the calculation and analysis,it can be seen that the overall performance of the model is good,the goodness of fit is as high as 93.49%,the error between the evaluation value and the real value is mostly concentrated below 10%.The random forest model has perfect applicability in the used car value evaluation.And through the analysis of the characteristic variable system,it can be seen that the new car price is the most important characteristic variable for the used car value,followed by the licensing time and driving mileage.In this paper,a model of used car value evaluation based on random forest is established from the theoretical point of view,which can not only reduce the evaluation cost and improve the evaluation efficiency,but also meet the needs of the developing market and provide a new way for used car value evaluation.Through the analysis of the evaluation results,it can be seen that the model can be applied to the vehicle evaluation of different brands,configurations and regions.And as long as the used car information is input into the model,the evaluation results can be obtained quickly.Therefore,the evaluation process is simple and easy to operate,and the model has certain use value.