Research On Influencing Factors Of Domestic Used Car Price Based On Hedonic Price Theory And Machine Learning Technology

In recent years,under the multiple influences of economic slowdown,travel mode upgrading,population structure adjustment and other factors,the trend of the domestic automobile market from the incremental stage to the stock stage has been basically determined.As the breakthrough direction of sustainable development in the stock stage of the automobile market,the used automobile industry has important research significance.The research on the influencing factors of used car price can improve the accuracy of price prediction of used car surplus value from consumers,promote the construction of used car scientific evaluation system,and accelerate the scale and standardization of used car industry.This paper,based on previous studies,constructs the influencing factor system of used car price from the characteristic price theory.Configuration factors,depreciation factors and brand factors constitute the individual characteristics of used cars.Policy factors,social factors and market factors constitute the extension characteristics of used cars.The analysis and verification of the influencing factors of used car price is transformed into the regression forecast of used car price.More than 70,000 pieces of used car data collected from used car trading website renrenche.com is deeply rearranged and then processed using multiple encoding methods data in combination with variable meanings.Taking used car related factors as independent variables and used car price as dependent variables,a variety of machine learning regression models are constructed.According to the model results and evaluation indicators,the influencing factors of used car price are verified and the model comparison is completed.The results show that the individual characteristics have a decisive effect on the used car price,and the epitaxial characteristics have a regulating effect on the used car price.Compared with the Classical Linear Regression Model,Decision Tree Regression,Random Forest Regression,Support Vector Regression,K-Nearest Neighbor Regression and Neural Network MLP Regression have improved performance in used car price prediction.Among them,Random Forest Regression Model has the best fitting effect on used car price under the premise of high computational efficiency,the results of feature importance of Random Forest Regression Model also show that the influence of configuration factors,depreciation factors and brand factors on used car price is gradually weakened,but in specific market segments,the influence of relevant factors on the used car price is different.