Araç Fiyat Tahmininde Makine Öğrenmesi Algoritmalarının Karşılaştırılması ve Performans Analizi
Comparison And Performance Analysis of Machine Learning Algorithms in Vehicle Price Prediction

Summary/Abstract: This study focuses on the problem of vehicle price prediction and aims to compare the performance of various machine learning algorithms while offering strategic insights for the industry. Linear Regression, Ridge Regression, Lasso Regression, Decision Trees, Random Forest, Gradient Boosting, Support Vector Regression (SVR), XGBoost, Multi-Layer Perceptron (MLP), and Radial Basis Function (RBF) models were tested on a dataset comprising 23.900 records collected between January 2022 and December 2023. The dataset includes technical specifications of vehicles—such as brand, model, engine size, and price-as well as customer-related features. During the data preprocessing phase, missing values were imputed, categorical variables were encoded, and outlier analysis was conducted. Lasso Regression emerged as the most successful model, achieving an R² value of 0,99 and low error rates. Random Forest, Gradient Boosting, and MLP also demonstrated strong performance with R² values around 0,97. In contrast, Decision Trees, SVR, and XGBoost showed weaker results, characterized by higher error rates and lower R² values. The findings highlight the critical importance of data preprocessing in model performance. Moreover, the integration of macroeconomic indicators and the application of hyperparameter optimization techniques have the potential to further enhance model accuracy and generalizability.

• Details
• Contents