Message-ID: <4B41E675.4050004@vanderbilt.edu>
Date: 2010-01-04T13:00:37Z
From: Frank E Harrell Jr
Subject: Are unpaired data suitable for DiagnosisMed's Diagnosis ?
In-Reply-To: <c895d7a51001040436p212d51bbr1570c5c32a1c601f@mail.gmail.com>
elaine kuo wrote:
> Dear,
>
> I wanna to compare AUC generated by two distribution models using the same
> sample.
>
> The AUC for model 1 consists of two columns, column A for 0/1 and column B
> for probability, eahc with the same row number of 3000.
> The AUC for model 2 consists of two columns, column A for 0/1 and column B
> for probability, eahc with the same row number of 10000 rows.
>
> I am wondering what value I should put in for gold and test in this
> function.
>
> Please kindly help and thank you.
>
> Elaine
Hmisc's improveProb (Pencina's methods) exists because comparisons of
AUC have low power. If you want to compare AUCs a more direct approach
is with rcorrp.cens in Hmisc.
You are confusing data for AUC which is a single summary measure. I
don't think you are using the term AUC in the right context.
Please consolidate your multiple queries into one note next time.
Frank
>
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
--
Frank E Harrell Jr Professor and Chairman School of Medicine
Department of Biostatistics Vanderbilt University