머신러닝을 활용한 중고 자동차의 가격 예측 모델에 관한 연구: 국내 브랜드를 중심으로

Abstract

The domestic used car market continues to grow along with the used car online platform. The used car online platform discloses vehicle specifications, accident history, inspection history, and detailed options to consumers. Most of the studies for predicting used car prices were studies using vehicle specifications. In addition, the relationship between the mileage and period of use of the vehicle and the used car price tended to appear nonlinear. In order to solve this problem, recent studies have attempted to reduce Cost Function using Machine Learning models. Accordingly, this study sequentially executed Regression based and Decision Tree based Machine Learning models using vehicle specifications and vehicle options. The implications of this study were, first, to make the most of their advantages by continuously executing two types of Machine Learning models. Second, which of the vehicle's specifications and option variables affects the price prediction of used cars, and the influence and direction of these variables were confirmed. This will help solve the problem caused by unfair information among used car sales officials.