Pemodelan untuk Pengiriman Barang dengan Memanfaatkan Jasa Kereta Apidi Jawa dan Sumatera

Abstract

Business on delivery sector has been growth recently, especially using railway. The development of infrastructures and facilities must be managed efficiently considering to an accurate information about the future need based on a historical data analysis. This article tried to develop models which can be used by decision makers to determine a reasonable plan to improve the profit. The Box-Jenkins method was employed to develop a model of historical data of goods distribution using train for Java and Sumatera area. The residuals was maintained using the GARCH model. This model was compared to the Trend and Seasonal Linear Model (TSLM) by evaluating the mean of absolute errors (MAE) and the root of mean squared errors (RMSE). The result shows that the ARIMA+GARCH model did better than ARIMA, SARIMA, and TSLM in predicting one and three month ahead for Java area, while the TSLM was more suitable for six and nine month forecasting than the others. On the other hand, the linear model provided the least values of MAE and RMSE for one, three, six, and nine month ahead prediction for Sumatera area