Hi there,
I have a question related to the sensitivity analysis in Vensim (I am using Vensim DSS DblPrec Version 5.3a). I have performed a sensitivity analysis varying 5 input parameters. I see the total effect on my outcome as confidence intervals, histogram etc. but how can I figure out the importance of different variables, which I have used for the sensitivity analysis, to the calculated outcome of my model i.e. to which variable is my output most sensitive?
Thank you very much for your help!
Sensitivity analysis
-
- Senior Member
- Posts: 1107
- Joined: Wed Mar 12, 2003 2:46 pm
The outcome of a model is the behavior over the time range of all variables in the model. I suspect you are interested in something much narrower in which case you might be better off simply running 10 simulations - up and down for each of the 5 variables - and looking at the results. The all constants sensitivity option in the optimizer might also be useful.
If you want to look at the correlation between input values and specific outcomes you can also export sensitivity results to a text file and use Excel or another application to computer correlation components. I believe there was a paper by Andrew Ford in the System Dynamics Review in the last couple of years that addresses this as well.
[Edited on 2007-11-5 by bob@vensim.com]
If you want to look at the correlation between input values and specific outcomes you can also export sensitivity results to a text file and use Excel or another application to computer correlation components. I believe there was a paper by Andrew Ford in the System Dynamics Review in the last couple of years that addresses this as well.
[Edited on 2007-11-5 by bob@vensim.com]