Forecasting the Casualty of Building Construction Using LS-SVM

Abstract: With the development of Chinese economy, the building industry has been becoming one of pillar industries. The enormous investment of engineering construction, together with its labor-intensive character, has made building safety a serious problem. Heavy casualties and serious property loss take place every year. In order to reduce the building accident and improve the management level of construction enterprise, we must forecast the casualty of building construction using some method. In view of the shortage of building safety data and the difficulty to collect them, we propose a new forecasting method based on support vector machine in this paper. We analyze some casualty data and construct a forecasting model with the method of support vector machine. The experiments prove that the method has advantages of lower error in simulation and higher precision in forecasting comparing with artificial neural network (Back propagation, BP). So it has a variety of application in the field.