08-02-2013 09:43 AM
I have read SAS paper on proc phreg. I am working on time to event model and my aim is to predict the time to event. Since the hazrad distribution is not known, I am using the Phreg.
1. Proc Phreg (cox regression) provides the cumulative survival probabilities and hazard ratio. So when I sort the data by ascending survival, I am not getting the lift the way it should be. The curve is below the baseline.
2. So is cumulative survival used to generate lift or any other statistic is computed to get lift.
I am clueless here and any help in this regard is greatly appreciated.