| Title |
Development of Machine Learning Predictive Model for Forecasting Demolition Waste Generation |
| Authors |
차기욱(Cha, Gi-Wook) ; 홍원화(Hong, Won-Hwa) |
| DOI |
https://doi.org/10.5659/JAIK.2023.39.2.229 |
| Keywords |
Waste Management; Construction & Demolition Waste; Machine Learning; Demolition Waste; Predictive Model; Artificial Neural Network; Support Vector Machine |
| Abstract |
Due to the rapid increase in Construction & Demolition (C&D) waste, C&D waste management (WM) management is an important challenge,
and Artificial Intelligence (AI) technology is being actively used as a smart WM strategy. Demolition waste (DW) predictive models were
developed and tested by applying artificial neural network (ANN) and support vector machine (SVM) based on a dataset consisting of
categorical input variables in this study. For this, DW predictive models with optimal performance were derived through hyper-parameter
tuning of ANN and SVM algorithms. As a result of this study, the predictive performance of the ANN and SVM models showed mean
absolute error (MAE) 71.730 and 79.437, root mean square error (RMSE) 119.414 and 104.979, coefficient of determination (R squared)
0.891 and 0.859 mean square error (MSE) 11020.556 and 14259.820 respectively. Therefore, the ANN model was confirmed to be a better
model for predicting the DW than the SVM model in this study. At this time, the mean of the observed values is 987.181 kg·m-2, and the
mean of the predictive values of the ANN and SVM models are 986.180 kg·m-2 and 991.050 kg·m-2, respectively. |