Application of Autoregressive Integreted Moving Average (ARIMA)-Kalman Filter Method in Predicting Value of Anggaran Pendapatan dan Belanja Daerah (APBD) North Sumatra Province

One method that can be combined with the ARIMA forecasting method is the Kalman Filter method, which is a method that can minimize error values ​​by estimating parameter values. This study aims to make predictions by applying the Kalman Filter method to the ARIMA model which is used to predict the value of the APBD of North Sumatra Province. In determining the ARIMA model to be used, first obtained stationary data. From the stationary data, several temporary ARIMA models will be identified through the ACF and PACF plots. Then these models will be tested whether they meet the assumptions of significance, normality, and white noise and have small AIC and SBC values. Furthermore, a simulation of APBD predictions is carried out for 6 years and the resulting error value is >50%. To minimize these errors, the Kalman Filter method is applied using polynomials of degree 1 and 2, where the error resulting from the application of the Kalman Filter method is <10%. From this study it is known that the use of the Kalman Filter method in the ARIMA model is very effective in producing good predictive values, because the combination of these two methods produces a relatively small error.