Abstract:
The autoregressive fractional integrated moving average model is a development of the
Autoregressive integrated moving average model which is used for data containing
elements of long memory. This model uses differencing decimal numbers. In this study,
the ARFIMA model is used for modeling data on the value of oil and gas and non-oil
exports in Indonesia with a monthly period from 2000 to 2020 and then forecasting for
the next 12 periods. The first step is to identify long memory using hurst statistics. The
result of the estimation of parameter d using the Rescaled Range Statistics (R/S) method,
the value of d=0.4 is obtained. The best ARFIMA model obtained is ARFIMA (2, d, 1)
with d=0.4 having an AIC value of 3457.114 and a MAPE value of 16.5%. This indicates
a good level of forecasting ability
