Prediction of pollutant emission characteristics in ISO50001 energy management in the Americas: Uni and multivariate machine learning approach

Abstract

The American continent is experiencing significant economic and industrial development driven by sustainability principles. In this context, discussions on improving energy consumption have become increasingly frequent and dynamic across various sectors of civil society, including the implementation of energy efficiency measures as advocated by the ISO50001 energy management standard. However, there is a pressing need to investigate which socioeconomic aspects are responsible for the issuance of this certification in the Americas and how these factors relate to characteristic industrial emissions, especially particulate matter. This study aims to evaluate the socioeconomic factors influencing ISO50001 standard issuance and how these adjusted factors correlate with particulate matter of 2.5 μm and 10 μm dimensions. To achieve this, machine learning techniques were employed, considering the complex nature and risk of data overfitting. Model fitting was performed through multiple lasso regression, and the relationship between the adjusted factors was examined through cross-correlation analysis. The analyses indicate a strong correlation of adjusted macroeconomic indicators, especially with PM2.5, suggesting an association with cardiorespiratory problems and methane-related origins. This work is of great relevance to academia as it proposes new concepts regarding the interaction between energy efficiency standards and particulate matter. For the industrial sector, the adjusted factors provide guidance for standard implementation while also helping to mitigate health issues. Additionally, for the government, these results can assist in formulating policies to address specific health problems related to this area.