Optimasi Jaringan Saraf Tiruan Menggunakan Algoritme Genetika untuk Peramalan Panjang Musim Hujan

Abstract

The focus of this research is to predict length of the rainy season in Indramayu using artificial neural networks (ANN) optimized by genetic algorithm (GA). The weights generated by the neural network is random, so the result of predictive value can change in each training. GA is a computational model that has operator selection, crossover, and mutation for generating a new population. The initialization of weights that is given by ANN will be optimized using GA. The predictors which are used in this research are the southern oscillation index (SOI) data and the beginning of rainy season (AMH) data with the length of rainy season data in previous year to predict the length of rainy season in the current year. The best result is obtained from the model on average region. On this region, the RMSE amounts to 14 days and the correlation coefficient between the observed values and predicted values amounts to 0.741 at 10% significance level, the RMSE amounts to 19 days and the correlation coefficient between the observed values and predicted values amounts to 0.694 at 5% significance level.