Penggunaan ARCH/GARCH dalam Penanganan Heteroskedastisitas Ragam Sisaan (Studi Kasus: Curah Hujan Bulanan Stasiun Kalijati)
MetadataShow full item record
Rainfall data collected by BMKG are classified as time series data because the data is observed in the same time interval. Some problems such as serial correlation, nonstationarity in variance and mean of the data, and heteroscedasticity in the residual of the ARIMA model are frequently found particularly in the fluctuated rainfall data. The nonstationarity of variance can cause variance heteroscedasticity in ARIMA model, therefore a more sufficient ARCH/GARCH model is needed to overcome this problem. The variance of the monthly rainfall data in Kalijati Station from 1991 to 2012 was not stationary. Therefore this study used ARCH/GARCH method to handle heteroscedasticity problem which was found in ARIMA model of the rainfall data. Modeling the residual variance using ARCH/GARCH method obtained a smaller MAD value than ARIMA modeling. ARIMA (0,1,1) x (0,1,1)12 – GARCH (1,1) was the best simultaneous model which did not contain heteroscedasticity in the residual variance. The ARIMA (0,1,1) x (0,1,1)12 – GARCH (1,1) model forecasted that the monthly rainfall average in 2013 year was 279 mm with the highest monthly rainfall occuring on January 2013, which was 454 mm.
- UT - Statistics