TIME SERIES MODEL FOR TRAIN PASSENGER FORECASTING

Abstract

Trains as a means of public transportation have an important role in connecting various regions of Jabodetabek. Therefore, it is necessary to have a deep understanding of the trend of train passenger movements and predict the number of train passengers in the next period in order to optimize the management and service of train passengers properly. In this study, we examine two methods that can be used as forecasting methods for train passenger data sourced from the Central Statistics Agency (BPS), namely ARIMA and Prophet. This study demonstrates that the optimal ARIMA model is ARIMA (0,2,1), achieving a Mean Absolute Percentage Error (MAPE) of 4.91% and a Root Mean Square Error (RMSE) of 1754.970. In addition, the Prophet model, which is an additive regression model designed by Facebook for time series forecasting was also obtained with a MAPE of 0.04% and an RMSE of 1170.59. Considering the MAPE and RMSE values of the two models, the Prophet model emerges as the most suitable for forecasting the number of train passengers in the Jabodetabek region.