Time series analysis is an observation that is built on time sequences. Time series analysis is useful in various fields, especially meteorology. One aspect of meteorology is rainfall, which can have an impact on human life. Rainfall has a complicated pattern to predict, so we need the best method for forecasting rainfall. There are several methods that can analyze the intensity of rainfall. Methods that can be used to predict rainfall are ARIMA method, Feed Forward Neural Network (FFNN) method, and hybrid ARIMA-NN. This study aims to obtain the best rainfall modeling and prediction based on the three methods above. The rainfall data used came from the Mini Weather Station (MWS) at Supiturang and Manggisari hamlets. Based on the results of the study, at Supiturang, the best model was ARIMA(1,1,1) with RMSE of 3.4326. At Manggisari, the best model is Hybrid ARIMA(1,1,1) FFNN(4-9-1) with RMSE of 3.1056.
Copyrights © 2021