| Over the past decade,the new energy vehicle trade has been booming.With the consumption of the first batch of new energy batteries put into the market reaching the critical point,the first batch of new energy vehicles ushered in the replacement period and naturally ushered in the second-hand market of new energy.The transaction proportion of new energy used cars in the second-hand car trading market is also increasing.The second-hand car trading market urgently needs to form a standardized and stable evaluation policy,which requires both the standardization of the early evaluation and testing process and the specialization of the later price evaluation process.However,at present,there is no complete and unified price evaluation standard in the new energy used car market,which will affect the development of the new energy used car market.This paper applies the machine learning method to the second-hand car price evaluation,explores the key factors affecting the second-hand price,and constructs a reasonable model to predict and evaluate the second-hand car price,so as to promote the healthy development of the new energy second-hand car market.Taking the official second-hand car of Weilai automobile as an example,according to the detection indicators provided by the official second-hand car,this paper uses the method of machine learning to predict and evaluate the price of second-hand car.In this paper,the internal used car data of Weilai automobile is used.Firstly,the missing value of the data is processed and transformed,and then the variance method,correlation coefficient method,feature selection method based on lasso regression,feature selection method based on decision tree and PCA principal component analysis are used to screen the important feature variables,and the important features are brought into the random forest,XGBoost and SVR model for prediction.Through the comparison of model effects,it is found that the prediction accuracy of the model based on feature selection is higher than that of PCA principal component analysis.Xgboost model has the best performance,with an average absolute error of only 8635 yuan.It is suitable to be used as a reference model for used car price prediction.Finally,according to the new energy second-hand car price evaluation model established in this paper,on the one hand,it can guide second-hand car businesses to establish the second-hand car price evaluation system,on the other hand,it can provide price guidance for consumers,so as to promote the development of second-hand car transactions to a more fair and open direction. |
