Prediction of waste diversion and identification of trends in construction and demolition waste data using data mining

Abstract

An excessive amount of construction and demolition waste is generated in construction projects daily. To avoid negative environmental externalities and to promote sustainability, it is necessary for the construction industry to improve its waste management practices at a project level. Analyzing large volumes of waste generation data created in construction projects can provide a valuable opportunity to extract actionable insights. In this research, construction projects were selected, and three different data mining techniques were applied with the objectives of (1) predicting the future volume of diverted waste based on the amount of waste generation of major materials; and (2) identifying trends of waste generation in different construction stages of the projects. Specifically, to achieve these objectives, this study applies multiple regression, artificial neural network, and clustering algorithms. Notably, this study contributes to the existing construction waste management body of knowledge by demonstrating the application of different data mining techniques to predict future waste diversion and extract actionable insights at the project level that help promote sustainability and waste reduction. Furthermore, this study allows industry practitioners to better manage construction waste by understanding patterns of waste generation according to different phases of the project, and predicting amounts of waste for diversion.