Analytic Method Of Uncertainties Of Government-invested Project Based On Subjective Bayesian Networks

Since the American financial crisis year 2008, global economy has slipped down. In attempting to maintain a high economic growth, Chinese government has invested a huge amount of its funds in infrastructure construction, which is approximately six hundred billion set for government-invested program every year. Government-invested projects are measures taken by the government to provide public service to people and improve their living quality. However, with such investments of a few years, issue on investment control has risen up. "Three over" issue is becoming serious, for which the major reason is the project decision period with the most uncertainties has a most effect on investment control. Analytical method for the uncertainties in current evaluation system cannot give an efficient analysis on the effects of various uncertain factors to economic outcome. Therefore, how to establish an efficient analytic method to these uncertain factors is the focus of this research.The article selected the government-invested program as study object, and explained the general analytic methods of uncertainties and its problems. Based on the study, we established a new analytic method for the uncertainties by combining structure analytic model and Bayesian networks. The study scope is as below:Firstly, we identified the uncertain factors that exert effects on government-invested project from literature study, selected key influential factors, categorized and analyzed the outcome. We found out that four influential factors that exert most effects on project investment are the accuracy of quantity estimation, day works labor rate, material rate and machine rate.Secondly, we applied ISM to analyze causality among the four factors and worked out their hierarchical structure, and built a complete structure-analytic model. We introduced Bayesian networks to the structure-analytic model to conducting ratio evaluation on uncertainties of amount of government-investment project. We defined nodes and their states in Bayesian model, and obtained initial ratio and conditional ratio of some nodes through questionnaire, and then applied software Netica for ratio calculation. It is a new model and solution algorithm for analysis on uncertainty.Finally, a case of new emergency and medical building of a city was introduced to evaluating and estimating amount of the investment, in contrast with sensitivity analysis, the conclusion proves the feasibility and effectiveness of Bayesian networks.