Probabilistic analysis of factors affecting highway construction costs: a belief network approach

Abstract

This paper presents the application of belief networks to make inferences in highway construction costs. The methodology is evolving; it works very well when sufficient information and incomplete quantitative data are available. It is an attempt to identify the extent of influence of selected variables on highway construction costs. Belief networks are an expressive graphical language for representing uncertain knowledge about causal and associational relations among construction cost variables. This then provides a graphical representation of probabilistic construction cost models. The graph-theoretic framework of the belief lends itself for modeling probabilistic dependence and flow information between different construction costs and related variables in overall highway construction cost determination.Key words: construction costs, belief networks, graphical models, uncertainty, and Bayesian network.