<![CDATA[This study aims to develop an enterprise forecasting information system for daily passenger demand prediction on the Yogyakarta–Solo Commuter Rail Line using Long Short-Term Memory. The proposed system integrates passenger demand forecasting and dashboard-based decision support to assist operational monitoring. The research used an experimental and system development approach. Daily passenger volume data were processed through data cleaning, chronological ordering, historical feature engineering, normalization, and sequence formation. Three LSTM feature scenarios were evaluated: historical features, historical and calendar features, and historical, calendar, and tourism event features. The forecasting model was then integrated into a web-based dashboard that provides historical trends, forecasting results, actual versus predicted visualization, model evaluation metrics, and downloadable prediction reports. The results show that the LSTM model with historical features achieved the best performance, with an MAE of 3,713.15, RMSE of 4,281.68, and MAPE of 14.57%. The system testing results indicate that the main dashboard functions run according to the expected outputs. Therefore, the proposed system can support data-driven passenger demand monitoring and operational decision-making for commuter rail services.]]>
Copyrights © 2026
