Data-driven prediction of construction and demolition waste generation using limited datasets in developing countries: an optimized extreme gradient boosting approach

An effective waste management plan must include reliable building waste generation data. This study seeks to build a trustworthy approach for predicting Construction and Demolition Waste (CDW) generation utilizing limited data. Different optimization algorithms are applied to eXtreme Gradient Boosting (XGBOOST) and compared to standard machine learning models including artificial neural network, support vector regression, and decision tree. The versatility and generalizability of the proposed approach are showcased by utilizing two distinct datasets in the Greater Bay Area of China and Tanta City of Egypt. The developed model reported excellent results, with testing \({R}^{2}\) $R^2$ values equal to 99.9. The research findings may be applied to the strategic development of waste management facilities, monitoring the urban metabolism, creating a circular economy using limited data. The study’s outcomes, as aligned with these 3R principles, lay the foundation for proactive waste management planning, resource optimization, and material valorization in the construction sector. This can potentially foster a more sustainable and circular approach to waste management, promoting a shift towards a more resource-efficient and environmentally conscious construction industry.