R help - Feb 2011 - My own distribution in survreg function

Hello,

I?m trying to do some analysis using survreg function.  I need to implement
there my own distribution with density: 

 lambda*exp(-lambda*y),  where     y = a1/(1+exp(-a2*x)). 
a1, a2 are unknown parameters and x >0. 

I need to get estimates of a1 and a2 (and lambda of course)
I?m really not good at programming.

Is there any way how to implement this distribution to survreg without ?big
programming skills??  Can anyone tell me how to enter unknown parameters
without giving them any exact initial value?

Any help would be greatly appreciated,
Katerina Konarikova
-- 
View this message in context:
http://r.789695.n4.nabble.com/My-own-distribution-in-survreg-function-tp3258645p3258645.html
Sent from the R help mailing list archive at Nabble.com.

-------  begin included message --------

I'm trying to do some analysis using survreg function.  I need to
implement
there my own distribution with density: 

 lambda*exp(-lambda*y),  where     y = a1/(1+exp(-a2*x)). 
a1, a2 are unknown parameters and x >0. 
--- end inclusion ------

 The survreg function can fit location/scale families where the location
parameter is Xb, X a matrix of predictors and b the parameters.  I don't
think that your distribution can be written in that form.
  An alternative is to take the apporach of the t-distribution for
survreg, which has the right for for fixed df.  Then wrap the survreg
function inside optim, using the latter to get an estimate of df.

No, I don't have time to give deeper programming help, due to other
projects due.

Terry Therneau