Predicting future prices of used light commercial vehicles with machine learning algorithms

In the current economic condition, second-hand vehicles have become a focal point for customers in almost every region of the world. In addition to normal customers, businesses with commercial purposes have also entered this circulation for their economic interests. The current values and the potential increase and decrease in the future values of the vehicles within this economic circulation have always intrigued customers, whether they are regular users or businesses. In this study, the second-hand prices of the commercial vehicles frequently used by businesses providing domestic and international logistics services have been aimed to estimate using machine learning algorithms that have recently become popular. In the study, machine learning algorithms such as Linear Regression, K-Nearest Neighbors, Decision Tree, and Random Forest have been employed for estimation. The dataset required for machine learning has been obtained through observation from two different websites featuring second-hand vehicle listings between 18.01.2023 and 05.06.2023. At the end of the specified period, data on 20.616 commercial vehicles, including information on 13 different criteria, has been collected for use in machine learning. As a result, the success rates of the machine learning algorithms used in the study have been compared using performance indicators such as the Correlation Coefficient (R2), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). In the last part of the study, by evaluating the findings obtained from the machine learning algorithms, recommendations for future research have been provided for both academia and researchers in the industry.