Statalist


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

Re: st: iweights and ROC curves


From   Steven Samuels <sjhsamuels@earthlink.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: iweights and ROC curves
Date   Thu, 11 Dec 2008 20:04:10 -0500

I should have made clear: to get the ROC Curve, use ordinary - logistic-, not -svy: logistic-. Mimic the -subpop- option with an - if- statement.

On Dec 11, 2008, at 7:30 PM, Steven Samuels wrote:

Do not use iweights.-lroc- after -logistic- with frequency weights works fine..

Austin Nichols once pointed out that you can compute fweights that give the same results as pweights to any number of decimal places:

For k decimal places, use
new frequency weight = round(10^k * probability weight,1)

So for 2 decimal places:

233.43 -> 23,343  a whole number.


-Steve


On Dec 11, 2008, at 1:30 PM, Patrick McCabe wrote:

Does any one know if the ROC curves produced by a logistic regression model using svy subpopulation and importance weights are 'legit'? The resulting model is the same as the model produced with pweights but if you use pweights lroc does not work.

*
*   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/

*
*   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   |   What's new   |   Site index