Big data grace: Implementations of the feature engineering and data science algorithms for environmental protection law

• Published in: Alexandria engineering journal Vol. 125; pp. 256 - 264
• Main Authors: Wu, Wenyue, Zhao, Yiming
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.06.2025; Elsevier
• Subjects: Big data; Data science algorithms; Environmental protection law; Feature engineering; Machine learning; Data science algorithms; Big data; Environmental protection law; Feature engineering; Machine learning
• ISSN: 1110-0168; 2090-2670
• DOI: 10.1016/j.aej.2025.03.121

This study is intended to predict CO2 emissions using a set of features. With this aim, three machine learning (ML) algorithms have been used, namely, support vector regression (SVR), Long Short Term Memory (LSTM), and multilayer perceptron (MLP). First of all, correlation analysis was performed which revealed a low level of multicollinearity among the set of features. Hereafter, moving towards the modeling and compared the ML models, the findings showed that SVR (Linear) is the most reliable one, showing superiority to the rest by having the least Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) values. On the contrary, the Dynamic LSTM model demonstrated the worst performance across all evaluated metrics. Specifically, it showed the highest values for MSE, RMSE, MAE, and MAPE. Static LSTM and SVR (RBF) models performed moderately, with Static LSTM marginally outperforming SVR (RBF) on the evaluation metrics like MAE and MSE. This will provide insight into guiding policy decisions in the future regarding strategies on environmental management and demographics development. This study highlights ML’s role in environmental monitoring, aiding policymakers with data-driven strategies to reduce CO2 emissions and shape sustainable policies.