Year: 2025
Type: article
Abstract: As the car industry has been developed rapidly, the used car market has also shown its potential due to the affordability and ability to make prediction on prices. This study aims to predict the prices of used car by using two methods: Linear regression model and LightGBM model and make comparison on the performance of two models. The dataset used are found in Kaggle which contain 5847 groups of data with 11 different variables in total affecting the price. In this paper, only 10 variables are chosen to process in the two models and evaluate the results. It has been found that LightGBM model are better than linear regression model with a higher efficiency, suitability and accuracy, with an R2 value of 0.962 for the training set and 0.930 for the test set, compared to Linear Regression's R2value of 0.700. Additionally, LightGBM demonstrates lower prediction errors (MAE: 1.103, MSE: 4.858, RMSE: 2.204) and better handling of large-scale data. To conclude, the LightGBM model has higher accuracy and is more suitable to predict the used car prices with higher efficiency especially when processing complex, large-scale data compared to linear regression model.
Source: Theoretical and Natural Science
Author: David QIAN
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4410418300