A Comparative Study of Linear Regression and Random Forest Models for Predicting Used Car Prices

This study deeply analyzed the problem of used car price prediction based on machine learning methods. By constructing two models, linear regression and random forest, and comparing their prediction performance, the essential influence of model structure on price prediction accuracy and generalization ability was explored. The study used public data sets for strict data preprocessing and feature engineering. The results showed that the random forest model was significantly better than linear regression in prediction accuracy, which was particularly prominent in the scatter plot of actual and predicted prices. At the same time, through the feature importance analysis of random forests, it was found that the number of engine cylinders and fuel type have a key impact on vehicle pricing, which further confirmed the ability of random forests to effectively capture nonlinear features. Although there is a certain skewness in the residual distribution of random forests, it is suggested that advanced models such as gradient boosting trees and external data can be further introduced in the future to improve prediction accuracy and robustness.