Analisis Hasil Prediksi Jumlah Penumpang pada Kereta Rel Listrik (KRL) di Jabodetabek Menggunakan Metode Arima

Abstract

The growing population of Jabodetabek has led to an increase in public transportation users, especially KRL. The increasing number of KRL passengers results in overcrowding which has an impact on service quality and passenger satisfaction. Previous studies have highlighted passenger dissatisfaction due to passenger crowding, emphasizing the need for adequate train carriages to reduce passenger crowding. Through the application of the Autoregressive Integrated Moving Average (ARIMA) method, future ridership can be forecasted based on previous data, allowing stakeholders to make informed decisions to improve service quality and address overcrowding issues in the Greater Jakarta area. By using three different models when forecasting, the evaluation matrix shows that the ARIMA (9,1,7) model with a division of 90% training data and 10% testing data successfully predicts the number of KRL passengers in Jabodetabek and has better results than the other two models. This is evident from the error value obtained through evaluation matrix testing, resulting in a low value with MAE of 52640.24, MAPE of 6.63%, and RMSE of 71955.90. So it can be concluded that the ARIMA (9, 1, 7) model with a division of 90% training data and 10% testing data can be used to predict the number of KRL passengers in Jabodetabek and provide quite good results.