Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: ROC curve analysis

From   Heather Gold <[email protected]>
To   [email protected]
Subject   Re: st: ROC curve analysis
Date   Mon, 30 May 2005 09:35:50 -0400

Of course, you're right, Ricardo. There seemed to be one "optimal" point when looking at the graph, and that's what point I wanted to identify. Thanks for the tip on detail option -- I'll try that.

Do you happen to know any good tutorials on ROC curve analysis, something beyond the basics?

At 07:44 PM 5/27/2005, you wrote:

Dear Heather,

What do you mean by optimal when using a ROC curve?
Rarely the highest sensitivity and highest specificity
occur at the same point, and in some cases the highest
sensitivity may occur at a point where the specificity
is very low. So even if you can display all the
possible cut point, you would still need to decide if
you rather have higher sensitivity or specificity.

That said, you can use the option �detail- option to
display the cut points.


--- Heather Gold <[email protected]> wrote:

> Dear List Members-
> I've run some -roctab- commands to estimate the area
> under the ROC curve,
> and I've also graphed the curves. I'm wondering if
> there's a way in Stata
> to find the optimal cutpoint on the ROC curve. I can
> visualize it to some
> degree, but is there a way to find the highest point
> on the curve, a way
> that's already programmed in Stata? I haven't found
> anything in the archives...
> Thanks for your help,
> Heather
> *
> *   For searches and help try:
> *
> *
> *

Ricardo Ovaldia, MS
Oklahoma City, OK

Do you Yahoo!?
Read only the mail you want - Yahoo! Mail SpamGuard.
*   For searches and help try:
Heather Taffet Gold, PhD
Assistant Professor
Department of Public Health
Weill Medical College of Cornell University
411 E. 69th Street
New York, New York 10021

E: [email protected]
P: (212) 746-1245
F: (212) 746-8544

* For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index