SAS/IML Software and Matrix Computations

Statistical programming, matrix languages, and more
BookmarkSubscribeRSS Feed
🔒 This topic is solved and locked. Need further help from the community? Please sign in and ask a new question.
Hanyu
Obsidian | Level 7

As is well known that quasi maximum likelihood estimation requires computing first derivative of the individual likelihood at each observation. Does anybody know how to implement it in SAS?d

1 ACCEPTED SOLUTION

Accepted Solutions
Rick_SAS
SAS Super FREQ

I assume you want to use finite difference approximations. Finite differences are available by using the NLPFDD function. See the article "Optimizing? Two hints for specifying derivatives."

View solution in original post

4 REPLIES 4
Rick_SAS
SAS Super FREQ

I assume you want to use finite difference approximations. Finite differences are available by using the NLPFDD function. See the article "Optimizing? Two hints for specifying derivatives."

Hanyu
Obsidian | Level 7

Thank you Dr.

Hanyu
Obsidian | Level 7

Dr. I have defined a module that returns a vector containing individual likelihood value at each observation. Then I want to compute the gradient using NLPFDD but SAS reports that the module must return  a scalar.

Rick_SAS
SAS Super FREQ

The module should return the likelihood value for a specified value of the parameters, given the data.  The entire data set is used as input to the likelihood function, so it is not correct to say that the likelihood is evaluated "at each observation."  Look at the article "Maximum likelihood estimation in SAS/IML" and see if that helps.  Notice that the module takes a single argument, shich is the vector of parameters fr the model.  The data are specified by using the GLOBAL clause.

sas-innovate-white.png

Join us for our biggest event of the year!

Four days of inspiring keynotes, product reveals, hands-on learning opportunities, deep-dive demos, and peer-led breakouts. Don't miss out, May 6-9, in Orlando, Florida.

 

View the full agenda.

Register now!

Discussion stats
  • 4 replies
  • 1525 views
  • 0 likes
  • 2 in conversation