| With the rapid development of China’s auto market,the used car market is expanding simultaneously.Vehicle residual value evaluation is an indispensable part of the market,it is widely used in used car transactions,automotive finance and other fields.However,China has not formed a unified and authoritative value evaluation standard yet,which is not conducive to the healthy development of the market.In the process of manual vehicle evaluation,the influence of subjective factors on the technical status and moral hazard is inevitable,evaluation result on a vehicle may vary with each individual or suffer from moral hazard.Besides,large batches of demands for residual value evaluation require lots of manpower and time.Based on text mining technology and topic model,this paper identifies the degree of subject bias in vehicle maintenance records,obtains the degree of bias in three themes of vehicle condition:repair,maintenance,and accident,and it is used as an important calculation factor of vehicle residual value evaluation;By sorting out the four methods on vehicle residual value assessment,selects the replacement cost method as the final evaluation method,and combines mileage and comprehensive analysis method to improve the residual value rate;in order to calculate the residual value rate,the impacts of vehicle residual value are analyzed,and a vehicle residual value evaluation system is constructed and AHP is used to calculate the weight of each level’s factor relative to the main target.The vehicle residual value evaluation model is constructed based on the replacement cost method and the improved residual value rate,residual value evaluation system,and subject model.This model calculates the adjustment coefficient through the vehicle’s driving life,vehicle condition,word of mouth and new vehicle sales,and combines the new vehicle price of the evaluated vehicle to obtain the final evaluation result.According to a large number of sample vehicle data,gets the residual value of them,and then compares the result with the data scrawled from "RenRenChe".The result shows that mean value of normal distribution based on relative error is-0.053,relative error of 88.75%of samples range between-0.2 and 0.2,the vehicle residual value evaluation model in this paper has satisfactory accuracy and practicality;While analyzing the residual value rate of sample vehicles form perspective of countries,brands,series and age,analysis indicates that:(1)The residual value rates of Japanese cars and German cars are relatively higher than other countries,which are 63.16%and 56.43%respectively,and that of domestic cars is 45.74%;(2)Japanese brands Honda,Toyota and Lexus are ranked at the top three in residual value rate,the rates of them are 74.94%、65、05%and 63.55%,and the rate of United States brand Buick ranked tenth;(3)the rate of Honda CR-V is the highest in all series,which is 83.12%,Honda Civic and Toyota Reiz are ranked second and third respectively,their rates of are 81.06%and 78.23%.(4)The residual value rate decreased with the increase of the vehicle using age,the average rate of vehicles aged 1-3 years is 65.47%,3-5 years is 54.94%,5-8 years is 45.75%,rate of vehicles over 8 years is 39.80%. |
