Sistem Prediksi Nilai Tukar Mata Uang Rupiah Berbasis Web Menggunakan Long Short Term Memory

Abstract

Economic, trade, and business relations between countries have become important aspects in the era of globalization and free trade today. There are several countries that have the largest import value of goods from Indonesia, including China, the United States, India, and Japan. In order for the process of international trade to take place, an exchange rate for each currency held by both countries is needed. However, because exchange rates are volatile, traders find it difficult to plan their trades. Therefore, this research aims to build a currency exchange rate prediction system for the four countries, namely Yuan, Dollar, Rupee, and Yen, using the LSTM method to assist international traders in conducting cross-border business. This study uses historical exchange rate data from each currency sourced from the Alpha Vantage platform for the period from January 2022 to December 2024, with closing prices as the target prediction data. The prediction model used is a model that has been built with hyperparameters sourced from previous research. The results of this study show that the model has a MAPE value of 0.004403% for the Yuan, 0.004965% for the Dollar, 0.005307% for the Rupee, and 0.013664% for the Yen. In addition, the model that has been built was successfully implemented in the form of a web application. The website development was carried out using one of the microframeworks of the Python programming language, namely Flask, and the library from the JavaScript programming language, namely ReactJS. The website that has been built. The evaluation results of the built system show that the system successfully executed all the test cases provided in the testing using the Black Box Testing method.