PEMODELAN DAN PREDIKSI CURAH HUJAN MENGGUNAKAN IMBAL HASIL HETEROSKEDASTIS DI KABUPATEN BANDUNG

Abstract

Bandung Regency located in West Java Province, Indonesia, has a tropical climate with high rainfall variation. This variable rainfall pattern is one of the main characteristics of tropical regions, where seasonal changes such as the rainy and dry seasons significantly impact annual rainfall patterns. In addition, high rainfall is often a trigger for natural disasters, such as floods and landslides, which can cause economic and social losses. Therefore, the ability to predict rainfall accurately is crucial, both for disaster risk mitigation and as a basis for decision-making in various sectors. This study aims to predict rainfall in Bandung Regency for the next year using a time series model approach, specifically a heteroscedastic model. The main focus of the study is to examine rainfall returns, or changes in rainfall over time while considering the nature of heteroscedasticity. In this condition, the variance of the data is not constant over time. The phenomenon of heteroscedasticity is common in rainfall data due to significant changes between seasons or extreme weather events. The study results indicate that the best model for predicting rainfall imbal hasil s is the Autoregressive Conditional Heteroskedasticity (ARCH) model. The selection of this model is based on the Akaike Information Criterion (AIC), which shows that the ARCH model outperforms the GARCH(1,1) model in handling variance instability in rainfall data. The advantage of the ARCH model lies in its ability to dynamically model variance instability, resulting in more accurate predictions. This model is not only capable of capturing high volatility patterns but also provides a more realistic estimate of large fluctuations that could occur in the future.

Keywords

• aic, arch model, garch model, mle, conditional variance.