Predicting Energy-Based CO2 Emissions in the United States Using Machine Learning: A Path Toward Mitigating Climate Change DOI Open Access

Longfei Tian,

Zhen Zhang, Zijun He

et al.

Sustainability, Journal Year: 2025, Volume and Issue: 17(7), P. 2843 - 2843

Published: March 23, 2025

Climate change is one of the most pressing global challenges that could potentially threaten ecosystems, human populations, and weather patterns over time. Impacts including rising sea levels soil salinization are caused by climate change, primarily driven activities such as fossil fuel combustion for energy production. The resulting greenhouse gas (GHG) emissions, particularly carbon dioxide (CO2) amplify effect accelerate warming, underscoring urgent need effective mitigation strategies. This study investigates performance outcomes various machine learning regression models predicting CO2 emissions. A comprehensive overview metrics, R2, mean absolute error, squared root-mean-squared cross-validation scores decision tree, random forest, multiple linear regression, k-nearest neighbors, gradient boosting, support vector was conducted. biggest source emissions coal (46.11%), followed natural (25.49%) electricity (26.70%). Random forest boosting both performed well, but had highest prediction accuracy among (R2 = 0.98 training, 0.99 testing). Support (SVR) neighbors (KNN) demonstrated lower accuracies, whereas tree displayed overfitting. were found to be extremely sensitive coal, gas, petroleum (transportation sector) based on sensitivity analysis. usage, KNN SVR maintained excellent R2 (0.94–0.98) less susceptible changes in variables. analysis provides insights into agreement discrepancies between predicted actual highlighting models’ effectiveness potential limitations.

Language: Английский

Predicting Energy-Based CO2 Emissions in the United States Using Machine Learning: A Path Toward Mitigating Climate Change DOI Open Access

Longfei Tian,

Zhen Zhang, Zijun He

et al.

Sustainability, Journal Year: 2025, Volume and Issue: 17(7), P. 2843 - 2843

Published: March 23, 2025

Climate change is one of the most pressing global challenges that could potentially threaten ecosystems, human populations, and weather patterns over time. Impacts including rising sea levels soil salinization are caused by climate change, primarily driven activities such as fossil fuel combustion for energy production. The resulting greenhouse gas (GHG) emissions, particularly carbon dioxide (CO2) amplify effect accelerate warming, underscoring urgent need effective mitigation strategies. This study investigates performance outcomes various machine learning regression models predicting CO2 emissions. A comprehensive overview metrics, R2, mean absolute error, squared root-mean-squared cross-validation scores decision tree, random forest, multiple linear regression, k-nearest neighbors, gradient boosting, support vector was conducted. biggest source emissions coal (46.11%), followed natural (25.49%) electricity (26.70%). Random forest boosting both performed well, but had highest prediction accuracy among (R2 = 0.98 training, 0.99 testing). Support (SVR) neighbors (KNN) demonstrated lower accuracies, whereas tree displayed overfitting. were found to be extremely sensitive coal, gas, petroleum (transportation sector) based on sensitivity analysis. usage, KNN SVR maintained excellent R2 (0.94–0.98) less susceptible changes in variables. analysis provides insights into agreement discrepancies between predicted actual highlighting models’ effectiveness potential limitations.

Language: Английский

Citations

0