Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Stata 11 discrim logistic leave-one-out


From   khigbee@stata.com
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Stata 11 discrim logistic leave-one-out
Date   Tue, 13 Apr 2010 08:42:16 -0500

Burkhard C. Schipper <bcschipper@ucdavis.edu> asks why there is
no -lootable- option with -discrim logistic- (and correspondingly
why there is no -estat classtable, loo- postestimation available
after -discrim logistic-) when it is available after the other
-discrim- commands.

The reason is that there are quick update formulas for performing
leave-one-out analysis for the other methods.

Think of the case where you have a lot of observations -- lets
say 10,000.  With -discrim lda-, -discrim qda-, and -discrim knn-
there are formulas that allow us to take the discrimination
results from the full sample and quickly alter it to remove one
observation.  We are unaware of quick update formulas for
logistic discrimination and so providing a -lootable- option for
-discrim logistic- was not provided since with moderate to large
size datasets, the computation time would be excessive.  With
10,000 observations we would need to run 10,000 logistic
regressions.

Ken Higbee
StataCorp
*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index