Ensemble Learning Model for Prediction of Natural Gas Spot Price Based on Least Squares Boosting Algorithm

Abstract

Among fossil fuels, natural gas has been considered a fuel of choice for increasing the supply of energy and reducing environmental contamination worldwide. Given that the supply chain of natural gas is characterized by imbalances between demand and supply on a timely basis, effective and fast models for accurate spot price prediction for natural gas would not only inform but also help various stakeholders to make wise decisions in the competitive natural gas marketplace. Proposed in this paper is a regression ensemble learning model based on least squares boosting algorithm for improving the prediction of natural gas spot price on a monthly basis. Ensemble learning models are becoming more preferred predictive tools due to their capability to train multiple weak learning models, and aggregate their outputs to produce a strong learning model with better overall accuracy. A broad scope of time series data from January 1997 to May 2020 from Henry Hub natural gas spot prices were employed in developing and testing the proposed model. Coefficient of determination (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), mean square error (MSE) and root mean square error (RMSE) are employed as performance metrics to evaluate the performance of the proposed model. Compared to four existing state-of-the-art approaches, overall results show that the proposed model exhibits better performance over the state-of-the-art methods with R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , MSE and RMSE scores of 0.9668, 0.3248, and 0.5699, respectively.