Risks in the construction sector evaluation with artificial neural network

This study, which is aimed at the assessment of risks in the construction sector with artificial neural networks in occupational health and safety, is a study on the modeling of artificial intelligence technology on accident data for the prevention of occupational accidents within the limits of occupational health and safety. The studies on the causes of occupational health and safety accidents related to this situation have been examined and the application areas of artificial intelligence technology have been evaluated, and a work accident predictability model study has been made with artificial neural networks. The data set used in the study was created from 644 construction accident data for the years 2016-2019 of the joint health and safety unit operating in the province of Istanbul. Accidents are grouped by entering in excel. By adding up the daily accidents that occurred during the month, the number of accidents for the month was obtained, 4 categorical values for the years 2016-2019 and 12 categorical values for the months were obtained. The data set consists of 2 independent variables as month and year, and the number of accidents 1 dependent variable. 48 new data sets were created from 644 accident data. In the study, 70% of the data set is reserved as training data and 30% as test data. The coefficient of determination of model performances was evaluated within the criteria of R2 and RMSE (Root Mean Square Error). With the model created with ANN, the input and output values, the training process by applying them to the network repeatedly, and the learning that predicts the number of accidents by minimizing the errors were realized.