ROC with more than one predictor

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

ROC with more than one predictor

Luigi
Dear all,
I would like to perform a receiver operating characteristics (ROC) curve
analysis on data obtained using molecular assays.

I have here an example: the first column is the sample number, the second is
the response variable and the last three columns are the predictors. I wish
to set the formula to test as R ~ P1 + P2 + P3 in order to obtain the
cut-offs for each of the predictors combined.

Any tip on how to proceed?

Thank you
Best regards,
Luigi

        R P1 P2 P3
1 0 0.725196908 0 0
2 0 1.089876198 0 0
3 0 1.487365838 0 0
4 0 6.214688703 3.625 0
5 0 48.26356415 13.4 0
6 1 9.494694175 0 0
7 1 15.57700003 4.666 0
8 1 45.53510205 15.57 0
9 1 0.373389928 12.55 1
10 1 0.692909116 0 1
11 1 15.05400384 3.45 1
12 1 20.83909961 6.9 1
13 1 23.24171357 0 1
14 1 43.51470584 0 1
15 1 49.80855616 3.55 1
16 1 664.0886249 10.37 1
17 2 60.85094536 3.725 0
18 2 62.7104185 0 1
19 2 143.7426283 6.5 1

______________________________________________
[hidden email] 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.
Reply | Threaded
Open this post in threaded view
|

Re: ROC with more than one predictor

Suzen, Mehmet
On Tue, Oct 23, 2012 at 11:52 PM, Luigi <[hidden email]> wrote:
>
> Any tip on how to proceed?

You may want to do check rocplus package. Its vignette is pretty good.
http://cran.r-project.org/web/packages/rocplus/vignettes/rocplus.pdf
In your case, I think pairwise comparison would be one approach


Best,
Mehmet

______________________________________________
[hidden email] 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.