Long Term Short Memory with Particle Swarm Optimization for Crude Oil Price Prediction

Prediction of crude oil prices is important in energy because of its significant impact on global markets and socioeconomic stability. Several machine learning methods have been employed to assist the industry. One of the popular neural network methods, Long Term Short Memory (LSTM), has demonstrate...

Full description

Bibliographic Details
Published in:ISAS 2023 - 7th International Symposium on Innovative Approaches in Smart Technologies, Proceedings
Main Author: Yusoff M.; Sharif M.Y.; Sallehud-Din M.T.M.
Format: Conference paper
Language:English
Published: Institute of Electrical and Electronics Engineers Inc. 2023
Online Access:https://www.scopus.com/inward/record.uri?eid=2-s2.0-85184809168&doi=10.1109%2fISAS60782.2023.10391535&partnerID=40&md5=64c2fa788eb3a8377c2bfc7da5133137
Description
Summary:Prediction of crude oil prices is important in energy because of its significant impact on global markets and socioeconomic stability. Several machine learning methods have been employed to assist the industry. One of the popular neural network methods, Long Term Short Memory (LSTM), has demonstrated its ability to predict time series data, particularly in predicting crude oil prices. However, the prediction performance still requires more research and evaluation to improve accuracy. This paper introduces a novel methodology that embeds Particle Swarm Optimization (PSO) into LSTM networks to minimize prediction errors. This study independently examines and identifies the most efficient hyperparameters within a ring topology of PSO, utilizing historical data on crude oil prices. The empirical findings indicate that the LSTM-PSO approach exhibits superior predictive accuracy compared to conventional LSTM networks and statistical time series techniques. The results suggest that the LSTM-PSO model demonstrates enhanced proficiency in capturing crude oil price patterns. © 2023 IEEE.
ISSN:
DOI:10.1109/ISAS60782.2023.10391535