<![CDATA[Accurate passenger demand forecasting is essential for optimizing the operational efficiency and sustainability of urban rail systems. This study aims to forecast monthly passenger numbers in the Palembang Light Rail Transit (LRT) system using optimized Exponential Smoothing (ES) models with parameter tuning for improved predictive accuracy. Three variants of the ES method are examined: Simple Exponential Smoothing (SES), Double Exponential Smoothing (DES), and Brown's Exponential Smoothing (BES). Furthermore, Seasonal ARIMA (SARIMA) is used as a benchmark to evaluate whether simpler ES models can match or outperform complex statistical approaches. Data from August 2018 to December 2023 are analyzed and split into training and testing sets (ratio 80:20). Model performance is evaluated using Mean Absolute Percentage Error (MAPE), Diebold–Mariano (DM), and Ljung–Box Q (LBQ) tests. The results show that the SES model with a smoothing parameter α = 0.9 achieves the best forecasting accuracy on the test data (MAPE = 8.6%), outperforming other ES variants and previous SARIMA-based models. These findings highlight that simpler ES-based models can effectively capture short-term transportation demand patterns in developing urban transit systems. Practically, the results of this study can provide valuable insights for LRT operators and municipal planners in designing responsive, data-driven operational strategies.]]>
Copyrights © 2025
