Variance based measures in sensitivity analysis (part I)
An understanding of how a computational model behaves in response to changes in its inputs is of fundamental importance to ensure a correct use of the models. This is provided by sensitivity analysis. This paper addresses the essence and role of sensitivity analysis. In addition it focuses on a class of sensitivity measures defined as global and quantitative. These are quantitative methods capable of decomposing quantitatively the variance of the model output according to the distribution functions of the input. Further, the paper focuses on a selected set of methods that, although different in name and implementation, are in fact estimates of the same underlying statistical quantity. This quantity is called a global sensitivity index.
Bibliographic Reference: Article: Statistical Science (1998)
Record Number: 199811397 / Last updated on: 1998-11-10
Original language: en
Available languages: en