search

neurodata / Multiscale-Network-Test

Testing independence between network topology and nodal attributes
Apache License 2.0
1 stars 1 forks source link

node contribution simulation #16

Closed jovo closed 7 years ago

jovo commented 7 years ago

(Xv,Zv) ~ MVN( mu, [s, Rv; Rv, s])

plot estimated contribution rank vs true rank

or binary, and plot something reasonable

youjin1207 commented 7 years ago

Maybe using absolute value of |c(v)| rather than c(v)? I do not have any intuition on relatively extreme negative value. Please refer to the appendix of MNT.

jovo commented 7 years ago
youjin1207 commented 7 years ago

@jovo @cshen6 contr.pdf

It looks like column sum of statistics matrix works as good as |both sum|. Since I have no clue how to interpret |both sum|, it would be better to use column sum. It seems that row sum could be significantly negative under dependence?

cshen6 commented 7 years ago

great! wonderful! that means we are right here!

On Thu, Nov 3, 2016 at 12:38 PM, Youjin Lee notifications@github.com wrote:

@jovo https://github.com/jovo @cshen6 https://github.com/cshen6 contr.pdf https://github.com/neurodata/Multiscale-Network-Test/files/569581/contr.pdf

It looks like column sum of statistics matrix works as good as |both sum|. Since I have no clue how to interpret |both sum|, it would be better to use column sum. It seems that row sum could be significantly negative under dependence?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/Multiscale-Network-Test/issues/16#issuecomment-258199039, or mute the thread https://github.com/notifications/unsubscribe-auth/ALX0y2hYrL-Uqvrv_egKNyxxyzHWWmrEks5q6g4hgaJpZM4KhVVc .

cshen6 commented 7 years ago

now we should really include the node contribution methodology / simulation / real data into section 2/3/4 respectively!

On Thu, Nov 3, 2016 at 12:49 PM, Cencheng Shen cshen6@jhu.edu wrote:

great! wonderful! that means we are right here!

On Thu, Nov 3, 2016 at 12:38 PM, Youjin Lee notifications@github.com wrote:

@jovo https://github.com/jovo @cshen6 https://github.com/cshen6 contr.pdf https://github.com/neurodata/Multiscale-Network-Test/files/569581/contr.pdf

It looks like column sum of statistics matrix works as good as |both sum|. Since I have no clue how to interpret |both sum|, it would be better to use column sum. It seems that row sum could be significantly negative under dependence?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/Multiscale-Network-Test/issues/16#issuecomment-258199039, or mute the thread https://github.com/notifications/unsubscribe-auth/ALX0y2hYrL-Uqvrv_egKNyxxyzHWWmrEks5q6g4hgaJpZM4KhVVc .

cshen6 commented 7 years ago

I think we should include node contribution to the projection description too as a sub-task.

Can you quickly make the same figure, but only keep red line and black line?

The legend will be "Independence Test" & "Node Contribution Test", y axis label just use "Power" title will be "MGC for Testing Independence and Node Contribution"

make sense?

On Thu, Nov 3, 2016 at 12:52 PM, Cencheng Shen cshen6@jhu.edu wrote:

now we should really include the node contribution methodology / simulation / real data into section 2/3/4 respectively!

On Thu, Nov 3, 2016 at 12:49 PM, Cencheng Shen cshen6@jhu.edu wrote:

great! wonderful! that means we are right here!

On Thu, Nov 3, 2016 at 12:38 PM, Youjin Lee notifications@github.com wrote:

@jovo https://github.com/jovo @cshen6 https://github.com/cshen6 contr.pdf https://github.com/neurodata/Multiscale-Network-Test/files/569581/contr.pdf

It looks like column sum of statistics matrix works as good as |both sum|. Since I have no clue how to interpret |both sum|, it would be better to use column sum. It seems that row sum could be significantly negative under dependence?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/Multiscale-Network-Test/issues/16#issuecomment-258199039, or mute the thread https://github.com/notifications/unsubscribe-auth/ALX0y2hYrL-Uqvrv_egKNyxxyzHWWmrEks5q6g4hgaJpZM4KhVVc .

jovo commented 7 years ago

agreed!

On Thu, Nov 3, 2016 at 1:02 PM, cshen6 notifications@github.com wrote:

I think we should include node contribution to the projection description too as a sub-task.

Can you quickly make the same figure, but only keep red line and black line?

The legend will be "Independence Test" & "Node Contribution Test", y axis label just use "Power" title will be "MGC for Testing Independence and Node Contribution"

make sense?

On Thu, Nov 3, 2016 at 12:52 PM, Cencheng Shen cshen6@jhu.edu wrote:

now we should really include the node contribution methodology / simulation / real data into section 2/3/4 respectively!

On Thu, Nov 3, 2016 at 12:49 PM, Cencheng Shen cshen6@jhu.edu wrote:

great! wonderful! that means we are right here!

On Thu, Nov 3, 2016 at 12:38 PM, Youjin Lee notifications@github.com wrote:

@jovo https://github.com/jovo @cshen6 https://github.com/cshen6 contr.pdf https://github.com/neurodata/Multiscale-Network-Test/files/ 569581/contr.pdf

It looks like column sum of statistics matrix works as good as |both sum|. Since I have no clue how to interpret |both sum|, it would be better to use column sum. It seems that row sum could be significantly negative under dependence?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/Multiscale-Network-Test/ issues/16#issuecomment-258199039, or mute the thread https://github.com/notifications/unsubscribe-auth/ALX0y2hYrL-Uqvrv_ egKNyxxyzHWWmrEks5q6g4hgaJpZM4KhVVc .

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/neurodata/Multiscale-Network-Test/issues/16#issuecomment-258206245, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjckJ22jPjJjuj_XIhCXNguMEWm5CBks5q6hOTgaJpZM4KhVVc .

the glass is all full: half water, half air. neurodata.io, jovo calendar https://calendar.google.com/calendar/embed?src=joshuav%40gmail.com&ctz=America/New_York