Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Journal Focus Action of Research Mathematic (Factor M)
Dawi, Herculianus Rowa
Unknown Affiliation
Education Mathematics Other

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Forecasting the Rupiah exchange rate against the US Dollar using the LSTM algorithm Multiyaningrum, Riska; Dawi, Herculianus Rowa; Hartanto, Raka Nurhaq Mulya; Haris, M. Al; Amri, Ihsan Fathoni
Journal Focus Action of Research Mathematic (Factor M) Vol. 8 No. 2 (2025): Vol. 8 No. 2 (2025)
Publisher : Universitas Islam Negeri (UIN) Syekh Wasil Kediri

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30762/f_m.v8i2.6530

Abstract

Exchange rates are a vital indicator of an economy's balance. The fluctuations of Indonesia's currency, the rupiah, against the USD influenced trade patterns, investment, and both monetary and fiscal policy. Exchange rate fluctuations affect international trade, investment, inflation, and overall economic stability. The high volatility of the Rupiah against the USD, driven by macroeconomic and monetary factors, has a significant impact on national economic policy, necessitating research that utilizes the latest data and adaptive models. To capture the nonlinear and complicated behavior of exchange rates, an advanced methodology for forecasting is needed. This journal utilizes the Long Short-Term Memory (LSTM) neural network model to forecast the exchange rate of the rupiah towards the dollar from March 1, 2022, up to February 28, 2025, in daily data. The data used in this research are sourced from www.bi.go.id, which provides the official daily exchange rate of USD to IDR. The Long Short-Term Memory method was chosen for modeling long-term dependencies within time series. After normalization, an 80/20 split is performed for training and testing on the dataset. The network runs optimization using three hidden layers with 50 neurons each and a batch size of 32 for 200 epochs. The optimal configuration, achieved through experimental trials, consisted of two hidden layers with 50 neurons, a batch size of 32, and 200 epochs. This is manifest in the fact that LSTM effectively captures movements in exchange rates, with an RMSE of 0.6226 and a MAPE of 0.3031%. This degree of accuracy enables the model to inform economic policy decisions based on data.
Co-Authors Haris, M. Al Hartanto, Raka Nurhaq Mulya Ihsan Fathoni Amri Multiyaningrum, Riska