Graphical Methods for Uncertainty and Sensitivity Analysis

Graphical methods play an important role in SA, since they provide a means of visualizing the relationships between the output and input factors. However, in complex problems, with many input variables, the problem of constructing useful graphical presentations becomes challenging, and the types of solutions become more varied. This chapter presents a number of familiar and perhaps not so familiar graphical tools which are used in SA. In chapter 2, Sampling methods, much use is made of the scatter plot for investigating pair-wise relationships and linearity, and here we present some modifications of such plots, which allow more than two variables to be studied. We are all familiar with the interpretation of the scatter plot, but in SA we are also interested on conditioning on the values of over input factors and on understanding the multi-dimensional relationships (very difficult to visualize, and only possible through considering different projections).