Difference between revisions of "Sensitivity analysis"

From Testiwiki
Jump to: navigation, search
(Adding references to glossary terms)
(content of Sensitivity analysis in Intarese wiki copied here)
Line 1: Line 1:
 
[[Category:Glossary term]]<section begin=glossary />
 
[[Category:Glossary term]]<section begin=glossary />
 
:A study performed to find out how the variation in the output of a model (numerical or otherwise) can be apportioned, qualitatively or quantitatively, to different sources of variation.<ref name="Sensitivity analysis">[[:en:Sensitivity analysis|Wikipedia definition on sensitivity analysis]]</ref><section end=glossary />
 
:A study performed to find out how the variation in the output of a model (numerical or otherwise) can be apportioned, qualitatively or quantitatively, to different sources of variation.<ref name="Sensitivity analysis">[[:en:Sensitivity analysis|Wikipedia definition on sensitivity analysis]]</ref><section end=glossary />
 +
 +
 +
==What is sensitivity analysis (SA)?==
 +
 +
'''Sensitivity analysis''' is a means of analysing the ways in which assessment outputs vary according to variation in assessment inputs. "Sensitivity" is the property of an assessment output that changes when the value or structure of an input is varied. Assessment output variability can be apportioned, qualitatively or quantitatively, to different sources of variation in the inputs (within variables or their location in the causal diagram etc.)
 +
 +
==How can SA be used in integrated environmental health impact assessment?==
 +
 +
Sensitivity analysis can be used in integrated environmental health impact assessment to determine which variables or steps in the full chain have the greatest effect (quantitatively or qualitatively) on the variation in the assessment's output. This allows the analyst to see clearly which aspects of a causal diagram are most liable to be affected by change in individual variables, in the structure of that model, and in the assumptions used to frame the assessment in the scoping phase.<ref>[http://sensitivity-analysis.jrc.it/default2.asp?page=sa A forum on sensitivity analysis. Joint Research Centre. Page accessed 13/11/2007]</ref> SA, then, serves as a step in the analysis of uncertainty crucial to generating robust and acceptable assessment results.
 +
 +
Sensitivity analysis can be used in dealing with:
 +
#Investigating the degree to which an assessment causal diagram resembles the process being modelled;
 +
#Judging the quality of model definition;
 +
#Establishing which factors contribute most significantly to variability in the output;
 +
#Interactions between factors
 +
 +
==Resources==
 +
 +
There are a number of online resources available in terms of both information on SA (including papers and tutorials), as well as software that can be used to carry out sensitivity analysis on integrated environmental health impact assessment models.
 +
 +
===Links===
 +
*[http://en.wikipedia.org/wiki/Sensitivity_analysis The Wikipedia article on sensitivity analysis]
 +
 +
===Software, tutorials and presentations===
 +
 +
*Carrying out uncertainty and sensitivity analysis in Analytica ([http://www.lumina.com/software/ch10.9.PDF External link to pdf])
 +
*[http://cran.r-project.org/src/contrib/Descriptions/sensitivity.html This package for the R environment can be used to perform sensitivity analysis] <small>External link</small> (The R software itself can be downloaded for free from [http://cran.r-project.org/ this link])
 +
*[http://sensitivity-analysis.jrc.it/ The JRC has produced this software (SimLab) and tutorials on the use of sensitivity analysis] <small>External link</small>
 +
*[http://library.lanl.gov/ccw/samo2004/ A set of proceedings for the 2004 4th International Conference on Sensitivity Analysis of Model Output] <small>External link</small>
 +
*[http://samo2007.chem.elte.hu/slides.html A set of presentations from the 2007 Fifth International Conference on Sensitivity Analysis of Model Output] <small>External link</small>
 +
*[http://sysdyn.clexchange.org/sdep/Roadmaps/RM8/D-4526-2.pdf An introduction to sensitivity analysis] <small>External link</small> (This introduction refers to the [http://www.iseesystems.com/softwares/Education/StellaSoftware.aspx STELLA software (proprietary)], although it is a useful general introduction to some basic concepts)
 +
*[http://msl1.mit.edu/rdn/d_table.pdf An introduction to sensitivity analysis in spreadsheet environments] <small>External link</small>
 +
*[http://www.modeling.uga.edu/tauc/background_material/Washington-Main.pdf The critique of modelling and sensitivity analysis in the scientific discourse. An overview of good practices. Transatlantic Uncertainty Colloquium (TAUC). Washington, October 10-11] <small>External link</small>
 +
*[[media:Andrea Saltelli.ppt|Presentation by Andrea Saltelli given at the EEA 2007 Uncertainty Training (INTARESE)]] <small>ppt 4.17MB</small>
  
 
==References==
 
==References==
 
 
<references/>
 
<references/>

Revision as of 12:24, 22 October 2009

<section begin=glossary />

A study performed to find out how the variation in the output of a model (numerical or otherwise) can be apportioned, qualitatively or quantitatively, to different sources of variation.[1]<section end=glossary />


What is sensitivity analysis (SA)?

Sensitivity analysis is a means of analysing the ways in which assessment outputs vary according to variation in assessment inputs. "Sensitivity" is the property of an assessment output that changes when the value or structure of an input is varied. Assessment output variability can be apportioned, qualitatively or quantitatively, to different sources of variation in the inputs (within variables or their location in the causal diagram etc.)

How can SA be used in integrated environmental health impact assessment?

Sensitivity analysis can be used in integrated environmental health impact assessment to determine which variables or steps in the full chain have the greatest effect (quantitatively or qualitatively) on the variation in the assessment's output. This allows the analyst to see clearly which aspects of a causal diagram are most liable to be affected by change in individual variables, in the structure of that model, and in the assumptions used to frame the assessment in the scoping phase.[2] SA, then, serves as a step in the analysis of uncertainty crucial to generating robust and acceptable assessment results.

Sensitivity analysis can be used in dealing with:

  1. Investigating the degree to which an assessment causal diagram resembles the process being modelled;
  2. Judging the quality of model definition;
  3. Establishing which factors contribute most significantly to variability in the output;
  4. Interactions between factors

Resources

There are a number of online resources available in terms of both information on SA (including papers and tutorials), as well as software that can be used to carry out sensitivity analysis on integrated environmental health impact assessment models.

Links

Software, tutorials and presentations

References