Artificial Neural Network Model for Forecasting Natural Disasters: Polak-Ribiere and Powell-Beale Comparison

Abstract The prediction problem is an interesting topic to be discussed today. many predictive methods used solve problems have become obstacle for researchers and academics alike. This study aimed analyze the ability of ANN method using Polak-Ribiere Powell-Beale conjugate gradients. dataset analysis disaster times-series data in Indonesia last ten years (2011-2020). Data obtained from Indonesian Disaster Geoportal sourced National Management Agency can seen on infographic menu website https://gis.bnpb.go.id/. results based that has been carried out, 4-10-1 architectural model with Conjugate gradient produce lower MSE Testing/Performance than method, another advantage faster time. And fewer iterations. So it concluded comparing these two methods, forecasting/predicting natural disasters because a better method.