Predicting the changes in the WTI crude oil price dynamics using machine learning models


Guliyev H., Mustafayev E.

RESOURCES POLICY, vol.77, 2022 (SSCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 77
  • Publication Date: 2022
  • Doi Number: 10.1016/j.resourpol.2022.102664
  • Journal Name: RESOURCES POLICY
  • Journal Indexes: Social Sciences Citation Index (SSCI), Scopus, International Bibliography of Social Sciences, Aerospace Database, Aquatic Science & Fisheries Abstracts (ASFA), Business Source Elite, Business Source Premier, Communication Abstracts, EconLit, Index Islamicus, INSPEC, Metadex, PAIS International, Pollution Abstracts, Public Affairs Index, Civil Engineering Abstracts
  • Keywords: Energy economics, Machine learning models, Crude oil price
  • Istanbul University Affiliated: No

Abstract

This study aims to use a monthly dataset from 1991 to 2021 to predict West Texas Intermediate (WTI) oil price dynamics using U.S. macroeconomic and financial factors, as well as a global crisis and crashes. We used advanced machine learning models such as Logistic Regression, Decision Tree, Random Forest, AdaBoost, and XgBoost in this study. According to the results, the XgBoost and Random Forest models outperform traditional models. We also used DeLong statistical test procedures to accurately compare machine learning models' per-formance. In addition, the study used SHAP -SHapley Additive exPlanations values to support model evaluation and interpretability. This new outline highlights the critical features of the WTI crude oil price prediction and provides appropriate model explanations by utilizing the practical SHAP values. The empirical findings showed that machine learning models could successfully and accurately predict the trend of WTI crude oil price changes. Our findings are important for policymakers, companies, and investors, as well as long-term energy-based economic development.