Modelling masonry crew productivity using two artificial neural network techniques

Abstract

Artificial neural networks have been effectively used in various civil engineering fields, including construction management and labour productivity. In this study, the performance of the feed forward neural network (FFNN) was compared with radial basis neural network (RBNN) in modelling the productivity of masonry crews. A variety of input factors were incorporated and analysed. Mean absolute percentage error (MAPE) and correlation coefficient (R) were used to evaluate model performance. Research results indicated that the neural computing techniques could be successfully employed in modelling crew productivity. It was also found that successful models could be developed with different combinations of input factors, and several of the models which excluded one or more input factors turned out to be better than the baseline models. Based on the MAPE values obtained for the models, the RBNN technique was found to be better than the FFNN technique, although both slightly overestimated the masons' productivity.