Error numerical analysis for result of rainfall prediction between Tsukamoto FIS and hybrid Tsukamoto FIS with GA

Abstract

Rainfall is one important aspect of everyday life, but now rainfall increasingly unpredictable. Therefore it needs to make an accurate method to predict rainfall with small error. Tsukamoto FIS and genetic algorithm is one of algorithms that can be used for prediction problems. Research using Tsukamoto FIS and hybrid Tsukamoto FIS with GA for forecasting rainfall had been done already. The prediction results generated from both methods have a diverse error value. Need an error analysis to determine which method is most optimal to predict rainfall with minimum error. Therefore, this study focuses on error numerical analysis on the result of rainfall prediction using Tsukamoto FIS and hybrid Tsukamoto FIS with GA. From the analysis, Tsukamoto FIS produce relatively small error, but this method is weak when predicting rainfall = 0 or no rain. While hybrid Tsukamoto FIS with GA produce small error for predicting rainfall = 0 or no rain. It concluded that a hybrid method Tsukamoto FIS with GA generate an error value more smaller than Tsukamoto FIS.