基于LightGBM的二手车价值评估的研究  

Research on Used Car Value Evaluation Based on LightGBM

作　　者：李钰 张蝶依 Li Yu;Zhang Dieyi(School of Information Engineering and Application,Hebei GEO University,Shijiazhuang 050031;Zhumadian Vocational and Technical College,Zhumadian 463000)

机构地区：[1]河北地质大学信息工程学院,石家庄050031 [2]驻马店职业技术学院,驻马店463000

出　　处：《现代计算机》2022年第20期65-67,共3页Modern Computer

摘　　要：2021年我国汽车保有量达到3.78亿辆,同时因为核心零部件的缺乏,使得二手车交易十分火热。但国内外关于二手车估价模型的研究较少且现有模型精度不够,使得在二手车交易中,没有可靠的价格为买卖双方提供参考。针对此问题,基于LightGBM提出一种二手车价值评估模型。模型首先对二手车数据进行格式转换,填补缺失值等预处理,然后对数据的不同特征进行相关性分析,最终使用LightGBM模型进行价格的评估预测,二手车价格预测的平均绝对误差比传统DNN、XGBoost模型降低了34.41%和11.83%,从而对二手车定价提出合理的建议。In 2021, the number of cars in China will reach 378 million, and the lack of core parts makes second-hand car trade very hot. However, there are few researches on second-hand car valuation models at home and abroad and the accuracy of existing models is not enough, so there is no reliable reference price for buyers and sellers to refer to in second-hand car trading. To solve this problem, a used car valuation model is proposed based on LightGBM. The model firstly preprocessed the second-hand car data by format conversion and filling in missing values, and then conducted correlation analysis on different features of the data.Finally, LightGBM model was used to evaluate and forecast the price of second-hand car. Compared with traditional DNN and XGBoost models, the average absolute error of second-hand car price prediction was reduced by 34.41% and 11.83%. Thus put forward reasonable suggestions on second-hand car pricing.

关 键 词：二手车 价格评估模型 LightGBM 

分 类 号：F224[经济管理—国民经济] F426.471