Skip to Main Content
Based on statistics learning theory, support vector machine method is a data driven model which can comprehensively evaluate the problem studied, automatically find out the correlation and hidden variables between various factors in the process of learning from previous sample data of subject studied, yet not need to explicitly give mathematical model of the problem. Factors affecting the implementation of construction project are uncertainty and complicated, and there also exists complicated nonlinear relations between these factors and the project risk results output. This paper firstly expound the project time risk and the basic theory of support vector machine for regression, then makes a SVM model to predict the time risk of a project to be built according to some previous similar projects risk information. We aim to make sure of the final project time before the project is implemented. Case study turns out that this method is feasible and reliable.
Date of Conference: 26-27 Dec. 2009