Dear statalisters,
Several searches through the statalist as well as googling a little around
on the keywords "mutual information" failed to provide a successful answer
to what I am looking for.
I would like to know if you are aware of any Stata module to compute the
normalized mutual informantion to score model quality (say, after a
logistic regression analysis).
Since this measure may have different names depending on the area of
research, here is a brief definition:
Normalized mutual information (NMI) is a measure of information
transmission based on Shannon's entropy. It has been
described by Forbes (1) as a performance measure for classifiers.
The NMI is the proportion of information provided by
the algorithm's classification, which is contained in the
true status or outcome.
(1) Forbes AD: Classification-algorithm evaluation: five performance
measures based on confusion matrices. J Clin Monit 1995,
11:189-206.
Thanks for any tip on additional fitness measures to examine model quality.
All the best,
Tiago
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
