Comparison of Predictions Foreign Currency Exchange Rates (USD, JPY, EUR, GBP, CHF, CAD) to Rupiah (IDR) using Artificial Neural Network and Long Short-Term Memory

Abstract

A country's currency exchange rate always moves, fluctuating erratically. Many researchers have tried to predict currency exchange rate movements to make informed decisions. We aim to determine the price prediction of foreign exchange rates against the Rupiah (IDR) and determine whether there are significant differences between the accuracy values of Artificial Neural Network (ANN) and Long Short-Term Memory (LSTM) methods for predicting exchange rates. It helps solve the problem of the instability of currency exchange rates over time. In this study, the steps began with collecting the data to use. Furthermore, transforming data and pre-processing data followed by performing data analytics using the ANN and LSTM. We compare the two methods to obtain the best result by measuring Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) values. We predict six foreign exchange rates (USD, JPY, EUR, GBP, CHF, and CAD) against Rupiah exchange rates (IDR) using historical data for 25 years, from January 01, 1996, to December 31, 2021, to get the best parameter based on the RMSE and MAPE evaluation results. From the parameters result, exchange rate prediction using LSTM gave the best RMSE and MAPE results based on the parameters used in each type of currency. Thus, it is concluded that LSTM is a better method for predicting currency exchange rates.