Rainfall Prediction Using Hybrid Adaptive Neuro Fuzzy Inference System (ANFIS) and Genetic Algorithm

Abstract

Tengger Indonesia is one of the rich areas in agricultural commodities and one of its commodities is potatoes. In the process of planting potatoes, rainfall data is used to determine the most appropriate planting time in order to harvest the maximum yield. However, the current rainy season is erratic and very difficult to predict the planting time, especially in the area of Tengger. It requires a method that can predict rainfall with the smallest error as possible. Adaptive Neuro-Fuzzy Inference System (ANFIS) is one of the prediction methods that are quite reliable because it is equipped with a network that can learn. The ANFIS uses Sugeno FIS in its architecture. To improve the prediction results, the Sugeno FIS will be optimized in boundaries of membership function and coefficient consequent rule before it goes into the process of training with ANFIS. A genetic algorithm is used for the optimization process. The results of rainfall prediction using hybrid ANFIS-GA are proven to produce smaller RMSE of rainfall prediction method that has never been done before.