Demonstration of a Spreadsheet-Based Tool for Uncertainty Propagation and Updating. K. W. Harrison, Environmental Systems Program, Department of Civil Engineering, Box 7908, North Carolina State University, NC 27695-7908
This paper will describe ongoing software development work at North Carolina State University and demonstrate the capability it provides Microsoft Excel users for modeling uncertainty in their existing spreadsheets. In addition to uncertainty propagation that is implemented by other spreadsheet-based uncertainty analysis software, this tool also updates uncertainty given observations of the stochastic or uncertain parameters. This updating is accomplished through application of the Bayesian Monte Carlo procedure that involves repeatedly sampling from the probability distributions of each of the model inputs prior to taking any observations. The probability distributions are then updated with the observations through application of Bayes’ rule. Use of this tool in conjunction with a decision model can help to determine whether further costly sampling is justified or whether there is sufficient information to make a decision in spite of the remaining uncertainty. Potential improvements to the Bayesian Monte Carlo method, including the use of importance sampling to improve the efficiency, will also be discussed. The use of this tool will be demonstrated using an illustrative application to an air pollution risk problem.