Lecture: Sensitivity analysis, an introduction
“Are the results from a particular model more sensitive to changes in the model and the methods used to estimate its parameters, or to changes in the data?”
This remark by Giandomenico Majone goes the heart of the problem setting of sensitivity analysis, a tool which all modellers from all fields of application use to improve the quality of their inference. Sensitivity Analysis is crucial both in the model construction and model interpretation phases, and is considered an important ingredient of model verification and validation.
Sensitivity analysis is the study of how the uncertainty in the output of a mathematical model or system (numerical or otherwise) can be apportioned to different sources of uncertainty in its inputs. A related practice is uncertainty analysis, which has a greater focus on uncertainty quantification and propagation of uncertainty; ideally, uncertainty and sensitivity analysis should be run in tandem (Source: https://en.wikipedia.org/wiki/Sensitivity_analysis)
The talk will review some principles of sensitivity analysis, good and bad practices, and some practitioners' insight on when to use what.
Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D. Saisana, M., Tarantola, S., 2008, Global Sensitivity Analysis. The Primer, John Wiley & Sons publishers.
Saltelli, A., Annoni, P., 2010, How to avoid a perfunctory sensitivity analysis, Environmental Modeling and Software, 25, 1508-1517.
Andrea Saltelli, Ksenia Aleksankina, William Becker, Pamela Fennell, Federico Ferretti, Niels Holst, Sushan Li, Qiongli Wu, 2018, Why So Many Published Sensitivity Analyses Are False. A Systematic Review of Sensitivity Analysis Practices, available on ArXiv.