Research on Residual Value Evaluation of In-use Vehicles Based on LightGBM Model

Abstract

There is no authoritative and complete evaluation system in traditional vehicle residual value evaluation. It requires high time cost and labor cost, and the result is inaccurate. In this paper, firstly, we establish a residual value evaluation indicator system covering vehicle information, market performance and macroeconomic factors. Secondly, an evaluation model based on LightGBM is presented. It has better performance in evaluation accuracy as compared to other evaluation models. Finally, a dynamic interval estimation technology is used to estimate the residual interval. This paper provides reference for the residual value related business of in-use vehicles, and promotes a more open and orderly automotive post-market environment.