Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.

# st: -predictnl- after -streg-

 From "Feiveson, Alan H. (JSC-SK311)" <[email protected]> To "[email protected]" <[email protected]> Subject st: -predictnl- after -streg- Date Wed, 11 Jul 2012 09:53:27 -0500

```Following Steve Samuels' suggestion, I tried -nlpredict- after running -streg-, but I get an error because the estimates of ancillary parameters (kappa and sigma) are involved. Apparently you still have to write out the "complicated" function, whether you use -nlcom- or -nlpredict- (see Steve's second posting on this).

Too good to be true.

Al Feiveson

. streg urplan,dist(gamma)

failure _d:  fail
analysis time _t:  S
exit on or before:  time .
id:  id

Gamma regression -- accelerated failure-time form

No. of subjects =          213                     Number of obs   =     11850
No. of failures =           24
Time at risk    =        11850
LR chi2(1)      =     16.19
Log likelihood  =   -63.468446                     Prob > chi2     =    0.0001

------------------------------------------------------------------------------
_t |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
urplan |  -1.180329   .2737132    -4.31   0.000    -1.716797   -.6438605
_cons |   6.323396   .4125432    15.33   0.000     5.514826    7.131966
-------------+----------------------------------------------------------------
/ln_sig |   .1162692   .3291606     0.35   0.724    -.5288737    .7614121
/kappa |  -.3319299   1.034859    -0.32   0.748    -2.360216    1.696356
-------------+----------------------------------------------------------------
sigma |   1.123298   .3697455                      .5892683    2.141298
------------------------------------------------------------------------------

. predictnl B = predict(surv),ci(Blow Bupp)
expression is a function of possibly stochastic quantities other than e(b)
r(498);

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```