A used car repricing method based on K-Means++ clustering and multiple linear regression

Abstract

A predictive pricing method based on k-means++ clustering and multiple linear regression is proposed to solve the unreasonable initial pricing of used cars and evaluate the model's low accuracy. To begin, all vehicles are divided into three categories based on the transaction period of the data sample. The K-Means++ clustering algorithm is then used to cluster the best-selling cars, and a multiple linear regression equation is fitted for each category as the regression equation for prediction and evaluation. Finally, the revaluation of unsold and unsalable vehicles is examined. In addition, two used car valuation methods, XGBoost and AdaBoost, are compared. Because the model performs well, the experiment shows that using the multiple linear regression method based on clustering to estimate car price is reasonable.