Penerapan Algoritma Bayesian Regulation untuk Estimasi Posisi Cadangan Devisa Indonesia

Foreign exchange reserves need to be predicted because it is a significant monetary indicator show the strength or weakness of country's economic fundamentals. Therefore, purpose this study estimate position Indonesia's foreign at end 2022 and 2023 so that government has benchmarks information in determining right policy remains stable. The estimation algorithm used Bayesian Regulation algorithm, one Artificial Neural Network algorithms. research data on (US$ million) obtained from reports Bank Indonesia. This will analyzed using three network architecture models 4-9-1, 4-18-1, 4-27-1. Based analysis used, results 4-27-1 model best lower Mean Square Error (MSE) value 0.00203297. Another benchmark seen more minor epochs (iterations) faster times than other two models, even though they both produce 100% accuracy rate. Thus, can concluded good enough model. prediction results, part slightly decreased compared 2021.