A comparative analysis of artificial intelligence techniques for carbon emission predictions in the construction industry

Abstract

The construction industry significantly contributes to global carbon emissions, necessitating urgent mitigation measures. This study addresses the challenge of predicting carbon emissions during construction projects using advanced artificial intelligence (AI) techniques. The performance of two AI models, Random Forests (RF) and Support Vector Machines (SVM), is compared to determine their effectiveness in forecasting emissions based on construction materials, techniques and project scale. Predictive models were developed using a dataset derived from previous research and real-world construction site data, ensuring accuracy through meticulous pre-processing, including data cleaning, normalization, and feature selection. The RF and SVM models were trained and tested on this dataset to evaluate their performance. The results show that the models achieve significant accuracy, and the RF model slightly outperforms the SVM in precision and reliability. This study underscores the potential of AI-driven approaches to improve sustainability in the construction industry. Insights from the analysis can inform industry stakeholders and policymakers in developing effective carbon reduction strategies, aligning with global efforts to combat climate change.