Forecasting the Exchange Rate of Yen to Rupiah Using the Long Short-Term Memory Method

Abstract

Long Short-Term Memory (LSTM) is a modification of the Recurrent Neural Network (RNN) designed to deal with the issues of exploding and vanishing gradients and makes it possible to manage long-term information. To tackle these problems, modifications were made to the RNN by providing memory cells that can store information for long periods. In this study, the objective was to forecast the exchange rate of Yen to Rupiah using the LSTM method. The data used in this research is daily purchasing rate data from January 2020 to May 2023 which consists of 848 observations. The data was divided into two sets: 80% for training and 20% for testing. For the forecasting process, experiments were conducted to identify the best model by adjusting several hyperparameters. The performance of each model was evaluated using the Mean Absolute Percentage Error (MAPE). Based on the experimental results, the best model obtained was the LSTM model with a batch size of 20, 150 epochs, and 50 neurons per layer, resulted in an MAPE value of 1,5399.