implement NMI measure

[cerebis]

implement or import NMI measure.

[koadman]

found this discussion of NMI behavior a while back:
https://arxiv.org/pdf/1110.2515.pdf

[cerebis]

Thanks for this reference. I thought I had it but nope.

This is exactly the sort of analysis that is really valuable to the field.
A similar discussion is involved in the development of the other measures
I've seen.

Seems we get lazy or the mechanism of information interchange regularly
fails. If someone becomes prominent from a good run of publications (this
can just be down to timing), we just take all their prior work as golden.

Need an analysis of that! Temporal study science references: "treating a
method as gold, beyond its used-by date; how long is too long?". There's
going to be an overlap, but there should be a egg-on-your-face™ point. :-)

On Fri, May 27, 2016 at 12:02 PM, Aaron Darling [email protected]
wrote:

found this discussion of NMI behavior a while back:
https://arxiv.org/pdf/1110.2515.pdf

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#23 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/AFunizNtHV7bGA2MpNRhFnYJJiGH7dBJks5qFlCygaJpZM4IoFtc
.

[cerebis]

https://github.com/aaronmcdaid/Overlapping-NMI