Forecasting Construction and Demolition Waste Using Gene Expression Programming

Abstract

Accurate forecasting of construction and demolition waste (CDW) generation could provide valuable information for the planning, design, and management of CDW at municipal levels. However, the lack of reliable forecasting approaches and historical records makes it difficult to predict the amount of CDW for a long- or short-term plan. To effectively tackle the CDW forecasting problem, a novel computer-based prediction model, gene expression programming (GEP), is introduced and tested. With the CDW and other data on predictor variables from the last two decades, the amount of CDW is forecasted in this study. Results and findings obtained from this research show that GEP is an effective model for predicting waste generation, with lower average forecasting error than the multiple linear model and the artificial neural network. Research issues related to model selection, training, and validation are also discussed in the paper.