Hi,

I ran a cox proportional hazard model to predict cumulative baseline hazard and xbeta. My code is as follows.

Code:
global time default_month
global event default_1
global xlist var1 var2 var3 var4 var5
global group auc0_pos1
stset $time,failure($event)
stcox $xlist, nohr
stcox $xlist
predict double xbeta, xb
predict ch, basechazard
I changed some variables, such as var4 var5, in the regression to log-transformed variables. Now the predicted xbeta and ch are so different and the interpretations of the predictions makes no sense. I'm wondering if it's still corrent to predict xbeta and ch using the predict function. If so, what changes in the code should I make? If not, how should I predict xbeta and ch? Thank you!