A Review of Machine Learning Models Used in Forecasting of Petroleum Products Prices

Abstract

Premium Motor Spirit (PMS) and other crude oil commodities play a pivotal role in shaping the nation's economy, influencing inflation, transportation costs, and overall economic stability. Accurate forecasting of these prices is essential for managing the inflation and ensuring economic stability. This study provides a systematic review of the application of machine learning models for predicting PMS and other crude oil commodity prices, focusing on literature from 2018 to 2024. The review identifies Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) networks as the most reliable models for forecasting PMS prices. For crude oil and natural gas, models such as Artificial Neural Networks (ANN) and Random Forest are more effective. Additionally, hybrid models that combine ARIMA with machine learning techniques offer improved accuracy by capturing both short-term fluctuations and long-term trends. The findings underscore that no single model consistently outperforms others across all commodities, highlighting the need for tailored approaches based on the specific characteristics of each commodity and dataset.