A few suggestions to add to Sensitivity2All-- my apologies if any of these exist:
1. It would be nice to save percentage differences for times other than the final time. For instance, if I'm interested in the effects of climate policy by 2030 and 2050 this feature would be handy!
2. It's currently very hard to save any of the output graphs. If I close the results window without savings, it seems I need to rerun sensitivity2all. It would be nice if I could repeatedly access that window without having to save it.
3. Allowing for variables to be changed by more than 10%. While the default feature is helpful because all parameters are changed by the same percentage, I may have different levels of uncertainty for each variable-- I may be confident my base value is within 10% of the true value for one parameter but only within 50% for another. This would help when analyzing the effect of empirical uncertainties in parameter values. Furthermore, in addition to the current percentage differences, it would be helpful to report the elasticities (percentage difference in outcome divided by percentage difference in parameters).
4. Finally, the definition of the integrated payoff is not in the documentation for sensitivity2all: https://www.vensim.com/documentation/se ... rnado.html. I know it's in the payoff documentation but I didn't realize the two were linked until I looked at the forum.
Sensitivity2All questions.
-
- Super Administrator
- Posts: 4751
- Joined: Wed Mar 05, 2003 3:10 am
Re: Sensitivity2All questions.
I've converted this to a post rather than suggestions for improvements.
I'll ticket this for a future release. It would be very easy to add in.akuptel wrote: ↑Wed Jan 22, 2025 6:29 pm A few suggestions to add to Sensitivity2All-- my apologies if any of these exist:
1. It would be nice to save percentage differences for times other than the final time. For instance, if I'm interested in the effects of climate policy by 2030 and 2050 this feature would be handy!
There is a Sens2All analysis tool. Right click the analysis toolbar and click "Customise toolbar". There is a Sens2All tool.
Select the constants in the Sens2All output window, right click and select "Sensitivity->Start sensitivity with selected constants".akuptel wrote: ↑Wed Jan 22, 2025 6:29 pm 3. Allowing for variables to be changed by more than 10%. While the default feature is helpful because all parameters are changed by the same percentage, I may have different levels of uncertainty for each variable-- I may be confident my base value is within 10% of the true value for one parameter but only within 50% for another. This would help when analyzing the effect of empirical uncertainties in parameter values. Furthermore, in addition to the current percentage differences, it would be helpful to report the elasticities (percentage difference in outcome divided by percentage difference in parameters).
Thanks, I'll update the documentation.akuptel wrote: ↑Wed Jan 22, 2025 6:29 pm 4. Finally, the definition of the integrated payoff is not in the documentation for sensitivity2all: https://www.vensim.com/documentation/se ... rnado.html. I know it's in the payoff documentation but I didn't realize the two were linked until I looked at the forum.
Advice to posters seeking help (it really helps us to help you)
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391
Units are important!
http://www.bbc.co.uk/news/magazine-27509559
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391
Units are important!
http://www.bbc.co.uk/news/magazine-27509559
Re: Sensitivity2All questions.
Great, thank you!
However, my third suggestion was that you could vary all the constants by a percentage other than 10% in Sensitivity2All. I know it's already possible to do this with the sensitivity analysis tool, but it would be way more convenient to do it in sensitivity2all so it could immediately output a tornado graph.
Also, another suggestion is to output elasticity, i.e., percentage change in output variables divided by percentage change in parameters. Obviously this can easily be done using the information already outputted, but it would just be more convenient if Vensim did this automatically.
However, my third suggestion was that you could vary all the constants by a percentage other than 10% in Sensitivity2All. I know it's already possible to do this with the sensitivity analysis tool, but it would be way more convenient to do it in sensitivity2all so it could immediately output a tornado graph.
Also, another suggestion is to output elasticity, i.e., percentage change in output variables divided by percentage change in parameters. Obviously this can easily be done using the information already outputted, but it would just be more convenient if Vensim did this automatically.