<![CDATA[The movement of used car prices is highly dynamic and subjective, making the manual estimation process prone to bias and financial risk. This study aims to develop a web-based used car purchase price estimation system using the Extreme Gradient Boosting (XGBoost) algorithm combined with a reverse calculation logic. The dataset was obtained from secondary market data and primary showroom transaction records, totaling 12,324 clean data after passing the Grouped-IQR outlier filtering process. The XGBoost model was optimized using Grid Search and validated through 10-Fold Cross-Validation. The results showed that the optimal model configuration achieved a Mean Absolute Percentage Error (MAPE) of 11.23%, a Root Mean Squared Error (RMSE) of Rp 54,779,437, and a Coefficient of Determination (R2) of 0.8386. This performance indicates a highly accurate forecasting capability. The model was successfully integrated into a Laravel-based web application via a Python REST API, allowing users to obtain fair market price predictions and maximum purchase bids to improve the efficiency and objectivity of decision-making.]]>
Copyrights © 2026
