Forecasting Bitcoin Price by Tuned Long Short Term Memory Model

Abstract

The interest for cryptocurrencies is high and hence this work focuses on providing a practical real-world application of the swarm metaheuristics and long short term memory model (LSTM). The goal is price forecasting which is interesting due to the high volatility of the cryptocurrencies. The authors apply LSTM for the solution of the problem which has been proven to reap results with this type of problem. The LSTM is further optimized by a swarm metaheuristic -arithmetic optimization algorithm (AOA). The solution was tested alongside familiar high-performing competitors with the use of standard metrics mean absolute error (MAE), mean squared error (MSE), mean absolute percentage error (MAPE), and root mean squared error (RMSE). These metrics have been used for comparison between the solutions, upon which the proposed solution obtained overall best performance that testifies to the improvement of the solution.

Decomposition aided attention-based recurrent neural networks for multistep ahead time-series forecasting of renewable power generation

“…These algorithms, and algorithms derived from their base have been applied in several fields with promising outcomes. Some noteworthy examples of metaheuristics applied to optimization problems include examples for crude oil price forecasting ( Jovanovic et al, 2022 ; Al-Qaness et al, 2022 ), Ethereum and Bitcoin prices predictions ( Stankovic et al, 2022b ; Milicevic et al, 2023 ; Petrovic et al, 2023 ; Gupta & Nalavade, 2022 ), industry 4.0 ( Jovanovic et al, 2023b ; Dobrojevic et al, 2023 ; Para, Del Ser & Nebro, 2022 ), medicine ( Zivkovic et al, 2022a ; Bezdan et al, 2022 ; Budimirovic et al, 2022 ; Stankovic et al, 2022a ), security ( Zivkovic et al, 2022b ; Savanović et al, 2023 ; Jovanovic et al, 2023c ; Zivkovic et al, 2022c ), cloud computing ( Thakur & Goraya, 2022 ; Mirmohseni, Tang & Javadpour, 2022 ; Bacanin et al, 2022d ; Zivkovic et al, 2021 ), and environmental sciences ( Jovanovic et al, 2023d ; Bacanin et al, 2022b ; Kiani et al, 2022 ).…”