Adaptive neuro-fuzzy inference system for Egyptian residential construction waste prediction

Construction waste (CW) significantly impacts environmental degradation, resource depletion, and construction costs, especially in rapidly developing regions. However, data on CW generation remain scarce in many developing countries, hindering effective waste management. This study presents an artificial intelligence (AI)-based predictive model to estimate CW quantities—specifically, concrete, bricks, and steel—using a case study in Egypt. Data were collected from 25 construction sites, incorporating variables such as total area, design consistency, worker experience, and waste reuse. The model employs an Adaptive Neuro-Fuzzy Inference System (ANFIS) to analyze the dynamic nature of construction activities and predict waste generation. Model accuracy was evaluated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the a20-index. The concrete model achieved R2 = 0.95, RMSE = 0.077, MAE = 0.071, a20 = 0.500; the bricks model showed R2 = 0.96, RMSE = 0.063, MAE = 0.043, a20 = 0.400; and the steel model resulted in R2 = 0.93, RMSE = 0.044, MAE = 0.042, a20 = 0.500. The results demonstrate the model's potential to support data-driven waste estimation where data is limited. These findings can inform strategies and policies for CW management by enabling the forecasting of waste volumes during construction phases.