<![CDATA[This research aims to predict used car prices and analyze the price determinants using an Explainable Machine Learning (XAI) approach. Used car price prediction presents a significant challenge in the automotive market, where pricing is influenced by various complex variables. The methodology involves comparing the performance of two machine learning models: linear regression (LR) and random forest (RF), trained on a dataset comprising 2,059 used car data points and 19 engineered features. The best-performing model is then interpreted using the SHAP (SHapley Additive exPlanations) method to identify the contribution of each feature. The evaluation results demonstrate that the Random Forest (RF) model exhibits superior performance compared to the Linear Regression model. The Random Forest model achieved a coefficient of determination (R2) of 0.819 and a Mean Absolute Error (MAE) of 294,591.0 . This performance is significantly better than the linear regression model, which yielded an R2 of 0.771 and an MAE of 716,221.3. The SHAP interpretive analysis identified the most significant price determinants. In sequential order, the five most dominant factors influencing price prediction are max power, car age, vehicle length (length_num), vehicle width (width_num), and kilometer (mileage). This finding provides transparent and justifiable insights into the key variables underlying price fluctuations in the used car market.]]>
Copyrights © 2025
