Machine learning applications for carbon emission estimation

In the context of escalating global climate change concerns, accurately estimating carbon emissions is crucial. This paper conducts a systematic literature review (SLR) on the application of machine learning (ML) techniques for estimating current and future carbon emissions. The study aims to evaluate the effectiveness of various ML algorithms across different sectors, identify sector-specific opportunities, and propose enhancements for ML-based carbon emission estimation.

The review highlights significant progress in the transportation sector, with notable research focusing on vehicle emissions. However, it identifies untapped potential in the energy and industrial sectors, where data accessibility and complexity pose challenges. The paper discusses the applicability of commonly used ML algorithms, including Artificial Neural Networks, Ensemble Methods, Support Vector Machines, and Extreme Learning Machines, emphasizing their strengths and limitations in different contexts. Key methodologies for improving ML performance in carbon emission estimation include hybrid modeling techniques, optimization algorithms, influential factor analysis, and data estimation methods. Despite advancements, challenges such as computational complexity, data quality, and model interpretability persist. The paper recommends enhancing optimization techniques, advancing predictor analysis, improving data collection practices, and focusing on sector-specific applications to address these issues.

By synthesizing existing knowledge and identifying critical research gaps, this study provides actionable insights to advance future research in ML-based carbon emission estimation. The main contribution of this work lies in its focus on practical aspects, rather than theoretical limitations of models, as emphasized in many existing studies. It highlights model performance in real-world scenarios, identifies key factors that restrict the efficient implementation of certain ML models in practice. Furthermore, the study presents a comprehensive guidance framework to provide an overview of the field and practical direction for application of machine learning in carbon emission estimation, paving the way for more effective real-world applications.