search

jhu-graphstat / LLG

0 stars 2 forks source link

Errorbars for Table 1 #6

Closed dpmcsuss closed 8 years ago

dpmcsuss commented 8 years ago

Jovo says

table 1 is what? average RE? where are errorbars? are they negligle?? say that or show them. are any of the differences significant? i wouldn't say anything is better or worse if it is in the noise. if we are testing, what test are we using? nothing is independent.

@TangRunze, do we have error bars for these numbers?

TangRunze commented 8 years ago

Table 1 is the average RE based on the dimension selected by ZG and USVT.

The motivation of Table 1 is: in Figure 4, although the square and triangle represent the average dimension selected by ZG and USVT, the corresponding MSE in the figure might be far away from the actual MSE based on ZG and USVT. So Figure 4: average over MSE based on the same dimension; choose the average of the dimension Table 1: at each replicate, we consider the MSE based on the dimension selected by ZG or USVT, then average over MSE.

I calculated the error bars, all the radius of the confidence intervals are less than 0.0075. I think the point is to say RE<1 when m is small rather than comparing which of ZG and USVT is better, so the result is good.

Do we want to say the CI is negligible in this case? Or list the CI or plot the error bars?

TangRunze commented 8 years ago

@dpmcsuss BTW, could you push all the pdf figures? Like scaled_RE.pdf, P_desikan.pdf, Abar_desikan_m5.pdf, Phat_desikan_m5.pdf, etc.

jovo commented 8 years ago

either stating they are negligble or plotting is fine. same with table. in both cases, however, errorbars ought to be mentioned...

On Thu, Aug 11, 2016 at 12:31 AM, Runze Tang notifications@github.com wrote:

@dpmcsuss https://github.com/dpmcsuss BTW, could you push all the pdf figures? Like scaled_RE.pdf, P_desikan.pdf, Abar_desikan_m5.pdf, Phat_desikan_m5.pdf, etc.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/jhu-graphstat/LLG/issues/6#issuecomment-239072121, or mute the thread https://github.com/notifications/unsubscribe-auth/AACjct8TCHSsp2XS4TevUgwdT_bAVIBoks5qeqWHgaJpZM4JhhPs .

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