Improving Crude Oil Price Prediction with CEEMDAN-Hilbert Transform and LSTM Models

Abstract

Crude oil price fluctuations significantly impact global supply chains, energy policies, investment decisions, and economic stability. Traditional economic models often struggle to predict these fluctuations accurately due to the market's complexity. This study explores the use of neural networks, specifically Long Short-Term Memory (LSTM) models, in crude oil price prediction. We integrate Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Hilbert transform for multiscale decomposition to enhance prediction accuracy. Comparative experiments demonstrate that the proposed method outperforms traditional models, including ARIMA and SVM, in key metrics such as MSE, RMSE, MAE, and R2. Despite the promising results, future research should address data limitations, optimize models, and explore applications in other financial domains to further improve predictive capabilities and practical relevance.