Sensitivity Analysis

Rating:
Summary: Sensitivity analysis for everybody
Review: Too often modellers do not realise that sensitivity analysis is an essential part of the model building process. This volume has a didactical value showing how SA is often useful - and sometimes essential- to complete the model building process and to interpret results properly. It guides the reader through an array of different approaches, illustrating in a generally clear fashion the specificity of the different techniques to different problem-setting.

Although this is a multi-authored book, the discourse flows clearly across (most of) the chapters and coveys the main element of this new discipline.

The authors-editors show an overall preference for sensitivity analysis methods capable of global quantitative sensitivity analysis; the sections of the book devoted to local methods and to regression analysis are rather a useful review than actually new material. The sections on variance-based methods and on high dimensional model representations are probably the most instructive for the educated reader.

The applications are in general well presented and instructive. These range from atmospheric chemistry to material physics. A chapter on available software is also offered. Finally the chapter from Beck and Chen (Assuring The Quality Of Models Designed For Predictive Tasks) establishes the needed link between the present raging debate on model validation and the use of adequate sensitivity analysis methods.