Comparison of Modelling Tools in Assessing Safety Performance of Construction Site

Abstract

Safety performance is the key factor in construction sites as it figures out the unsafe act, condition and supervision of the worksite. Three different modelling tools such as Artificial Neural Network (ANN), Genetic Programming (GP) and Multi-linear Regression (MLR) are used to assess the performance of the site. Furthermore, the performance of these models is also analysed with respect to various conditions to assess its suitability. Results indicates that all the three models can be used to predict the safety performance but GP based models have higher flexibility in extrapolating the results outside the training range with highest accuracy. Thus, it is recommended to employ GP based models in predicting safety performance in construction sites.