Creating a cutting-edge neurocomputing model with high precision

Abstract

Abstract The prediction of oil prices has a significant impact on the economies of countries, particularly in oil-rich nations like Iraq, and affects the labor market. Prediction techniques are vital tools for extracting knowledge from complex databases, such as oil prices. This study aims to develop a prediction model that accurately determines oil prices based on seven fundamental characteristics, including Date, WTI, GOLD, SP 500, US DOLLAR INDEX, US 10YR BOND, and DJU. The proposed model utilizes advanced neurocomputing techniques that analyze the seven features over a ten-year period. The model comprises three main stages: preprocessing, determining feature importance through computing correlation, entropy, and information gain, and splitting the dataset into training and testing. The first part of the dataset builds the predictor called Hybrid Model to Oil Price based on Neurocomputing Techniques, while the second part evaluates model using three error measures: R2, MSE, and MAE. The model proves its ability to provide accurate predictions with low error rates. Multivariate analysis shows that WTI, GOLD, and US DOLLAR INDEX have a more significant impact on oil prices, with information gain values of WTI = 11.272, GOLD = 11.227, and DJU = 11.614. The Gate Recurrent Unit neurocomputing technique demonstrates its ability to handle datasets with features that behave differently over multiple years and provides accurate predictions with low errors in a short time, withR2 = 0.945, MSE = 0.0505, and MAE = 0.1948. This study provides valuable insights into the prediction of oil prices and highlights the efficacy of advanced neurocomputing techniques for extracting knowledge from complex databases.

Keywords

• Network of oil
• GRU network
• Forecasting
• Advanced neurocomputing