Thats great thanks,
I suppose it is hard to move away from a more "traditional" measure of
performance such a percentage correct, at least for the relatively amateur
statisticians among us who have been graded on such a system.
The difficulty comes in reporting the effectiveness of the model to
Thats great thanks
I guess it is hard to not use % as a performance measure when that is what is
commonly used in everyday life.
So when i come to predicting the response of new data ( using the estimated
mean Y ) which i am more comfortable with i can say -
Species A - 2.12 - Therefore this i
% correct is an improper scoring rule and a discontinuous one to boot. So it
will not always agree with more proper scoring rules.
When you have a more difficult task, e.g., discriminating more categories,
indexes such as the generalized c-index that utilize all the categories will
recognize the
Thanks Frank,
I have one small question regarding this, understand you are very busy and if
you cant answer i would greatly appreciate any thoughts from the list.
> Split-sample validation is not reliable unless you have say 10,000 samples to
> begin with
I am a little confused. When i ran the
Thanks Peter. I think you're right.
Frank
-
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context:
http://r.789695.n4.nabble.com/predict-lrm-Design-package-tp2546894p2547146.html
Sent from the R help mailing list archive at Nabble.com.
__
A few comments; sorry I don't have time for any more.
- Combining categories is almost always a bad idea
- It can be harder to discriminate more categories but that's only because the
task is more difficult
- Split-sample validation is not reliable unless you have say 10,000 samples to
begin wit
On Sep 20, 2010, at 15:50 , Frank Harrell wrote:
>
> You sent a private note about this which I just took the time to answer.
> Please send only one note, and please post my reply to you to r-help.
I have been annoyed by this at times as well. However, I have come to suspect
that it is actual
A few comments; sorry I don't have time for any more.
- Combining categories is almost always a bad idea
- It can be harder to discriminate more categories but that's only because the
task is more difficult
- Split-sample validation is not reliable unless you have say 10,000 samples to
begin w
You sent a private note about this which I just took the time to answer.
Please send only one note, and please post my reply to you to r-help.
Frank
-
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context:
http://r.789695.n4.nabble.com/predict-lr
9 matches
Mail list logo
