IJCCS (Indonesian Journal of Computing and Cybernetics Systems) Vol.17, No.1, January 2023, pp. 33~44 ISSN (print): 1978-1520, ISSN (online): 2460-7258 DOI: https://doi.org/10.22146/ijccs.80776 ◼ 33 Received December 30 th ,2022; Revised January 19 th , 2023; Accepted January 30 th , 2023 Improved LSTM Method for Predicting Cryptocurrency Price Using Short-Term Data Risna Sari 1 , Kusrini Kusrini* 2 , Tonny Hidayat 3 , Theofanis Orphanoudakis 4 1,2,3 Dept. of Information Technology, Universitas Amikom Yogyakarta, Indonesia 4 School of Science and Technology, Hellenic Open University, Patras, Greece e-mail: 1 risnasari@students.amikom.ac.id, * 2 kusrini@amikom.ac.id, 3 tonny@amikom.ac.id , 4 fanis@eap.gr Abstrak Seiring dengan berkembangnya cryptocurrency, tidak dipungkiri bahwa harga dari crypto tidaklah stabil. Salah satu faktor yang mempengaruhinya yaitu meningkatnya volume transaksi yang menarik minat peneliti untuk melakukan penelitian dalam mengembangkan metode prediksi harga koin dari cryptocurrency. Hasil prediksi dipengaruhi oleh penggunaan metode, algoritma hingga jumlah data. Pada penelitian ini akan dilakukan pemodelan prediksi dengan menggunakan metode LSTM dan data jangka pendek. Penelitian ini akan melakukan 2 percoba menggunakan metode LSTM sederhana dan memanfaatkan multivariate time series dengan LSTM. Hasilnya diperoleh nilai terkecil prediksi menggunakan skenario pembangian alokasi data 80/20, inputan layer LSTM = 360, Epoch = 500 yang dilakukan yaitu koin Solana dengan RMSE = 0.111, R2 = 0.9962. Dapat disimpulkan bahwa penggunaan data jangka pendek dapat digunakan dalam pembuatan pemodelan prediktif, namun perhatian khusus perlu diberikan pada karakteristik dataset yang digunakan dan metodologi pemodelan, dan diharapkan hasil penelitian ini dapat dapat dimanfaatkan dalam penelitian selanjutnya Kata kunci—Cryptocurrency, Prediksi, Long short-term memory (LSTM), Short-term data Abstract As cryptocurrencies develop, it cannot be denied that crypto prices are volatile. One of the influencing factors is the increasing volume of transactions which attracts the interest of researchers to conduct research in developing coin price predictions from cryptocurrencies. The method, algorithm and amount of data affect the prediction results. In this study, prediction modelling will be carried out using the LSTM method and short-term data. This study will conduct two experiments using the simple LSTM method and utilising multivariate time series with LSTM. The smallest predicted value is obtained using an 80/20 data allocation distribution scenario, input layer LSTM = 360, Epoch = 500, a Solana coin with RMSE = 0.111, R2 = 0.9962. It can be interpreted that short-term data can be used in making predictive models. Still, special attention needs to be paid to the characteristics of the dataset used and the modelling methodology, and it is hoped that the results of this study can be used in further research. Keywords— Cryptocurrency, Prediction, Long Short-term Memory (LSTM), Short-term data 1. INTRODUCTION The emergence of the Industry 4.0 era in the technological realm has affected all aspects of human life. One can feel the presence of a revolution impacting a consumer society with social, cultural, and economic changes. IoT-based systems, Robotics, and Cloud systems contribute to the realisation of the Industry 4.0 revolution and a new digital economy following this revolution. The digital economy that is felt in everyday life is characterised by transaction processes that are carried out not only traditionally but also digitally processes, also known as digital payments. Traditionally implemented financial transaction systems rely on third parties,