- Notifications
You must be signed in to change notification settings - Fork253
Group analysis support#358
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to ourterms of service andprivacy statement. We’ll occasionally send you account related emails.
Already on GitHub?Sign in to your account
base:experimental-methods
Are you sure you want to change the base?
Group analysis support#358
Uh oh!
There was an error while loading.Please reload this page.
Conversation
Raise warning if method does not support groups
lbteixeira commentedSep 20, 2020
Hello,@ConnectedSystems. Sure, it would be great to help, but this week I'm a little too busy with my PhD thesis. I can work on this in about 1 week, if it's not too late. |
Thanks,@lbteixeira There's no deadline and so. no rush. I'll come back to this myself in a week or so as well. |
One thing to consider with grouping approaches is that you can get cancellation effects. Morris uses an average of absolute elementary effect when computing sensitivity indices for groups, rather than the average. For example with two parameters in one group, if one parameter has a strong positive effect, and another a strong negative effect, these will cancel out when averaged and the group will seem to have a negligible effect. |
Uh oh!
There was an error while loading.Please reload this page.
Support analysis of parameter groups, or raise warning when they are not supported.
The method implemented just generalizes the existing approach for Morris - averaging the effect across group members.
With apologies to@lbteixeira for replacing the recently added
_compute_grouped_metric()DGSM and HDMR implementations were a little over my head so could not adjust the code to allow group analysis at this time (perhaps@sahin-abdullah or@lbteixeira would be willing to help here?).