Penerapan Metode ARIMA Ensembel pada Peramalan

(Studi Kasus: Inflasi di Indonesia)

Abstract

ARIMA ensemble is a method of combination forecast results from multiple ARIMA models. ARIMA method as individuals and ARIMA ensemble as a combination model to forecasting of national inflation in Indonesia. Ensemble method used to combine the forecast result in this study were averaging and stacking. The data used in this study is the nasional monthly inflation of Indonesian from January 2010 to December 2014. The results showed that for forecasting the next twelve months ensemble averaging method produces the smalles RMSE values ​​and obtained models equation where z_t⁽¹⁾ is ARIMA models (2,0,2) and z_t² is ARIMA models (2,0,3). Based on ARIMA ensemble averaging model the monthly inflation forecasting national Indonesia next twelve months forwards experience of fluctuation where highest inflation in January 2015, that is 1,13% and smallest in March 2015, that is equal to -0,13%.