turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Scoring proc phreg

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-06-2011 11:43 AM

Hello, this is my first time using this website. I am using proc phreg and outest function to build and output a model. Then I am using proc score to score the model. The scored observations are between 3 and 11 (as opposed to a probability between 0 and 1). What does this represent? Is there a formula to calculate probability of survival from these numbers? Thank you in advance for your help!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

09-06-2011 06:15 PM

I think I can answer the first part, but not the second at this moment.

Your getting the linear predictor, ie the xbeta.

You can test this using the following code and the sample data from proc phreg provided in the SAS documentation:

proc phreg data=Myeloma outest=estimates;

model Time*VStatus(0)=LogBUN HGB SCalc;

output out=estimates2 xbeta=sample1 ;

run;

data sample;

set Myeloma;

drop time;

if mod(_n_,2)=1 then output;

run;

*Score the data;

proc score data=sample out=results score=estimates TYPE=PARMS;

var logbun hgb scalc;

run;

The xbeta (sample1) should be the same as the numbers in the results table.

Proc SCORE seems to only multiply the parameters by the values, ie matrix math, so you'd need the linear estimators out of the proc Phreg which at this point I can't even recall is possible (just back from vacation).

Also wouldn't the probablilty always be 1 (ie always die or answer a call if waiting long enough?) so you might need to clarify what you're looking for as well. Isn't that an assumption of survival analysis, without censoring present?