Re: [R] Why different results with different initial values for MLE(optim)!

From: Ravi Varadhan <rvaradhan_at_jhmi.edu>
Date: Fri 23 Jun 2006 - 02:44:39 EST


It is difficult to say much since you haven't described your likelihood function. However, there are some general things that you can do: (1) make sure that you are "maximizing", by defining your objective function as the negative of likelihood, (2) use log-likelihood rather than the likelihood function, (3) look at the value of the objective function at different converged points to see whether you truly have multiple local maxima or if the likelihood is very flat near the optimum, (4) look at convergence criterion from the output to see whether you have true convergence, and (5) try different optimization techniques in "optim" or try "fitdistr" function in MASS library.

Ravi.



Ravi Varadhan, Ph.D.
Assistant Professor, The Center on Aging and Health Division of Geriatric Medicine and Gerontology Johns Hopkins University
Ph: (410) 502-2619
Fax: (410) 614-9625
Email: rvaradhan@jhmi.edu
Webpage: http://www.jhsph.edu/agingandhealth/People/Faculty/Varadhan.html

> -----Original Message-----
> From: r-help-bounces@stat.math.ethz.ch [mailto:r-help-
> bounces@stat.math.ethz.ch] On Behalf Of Xin
> Sent: Thursday, June 22, 2006 12:10 PM
> To: r-help@stat.math.ethz.ch
> Subject: [R] Why different results with different initial values for
> MLE(optim)!
>
> Hi, All:
>
> I used optim() to minimise likelihood function for fitting the data to
> a partiuclar distribution. The function is converged and the value of log-
> likelihood is different when I change the intial value.
>
> Whether it means the program does not work well?
>
> Thanks!
>
> Xin
> [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help@stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-
> guide.html


R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Fri Jun 23 02:48:36 2006

Archive maintained by Robert King, hosted by the discipline of statistics at the University of Newcastle, Australia.
Archive generated by hypermail 2.1.8, at Fri 23 Jun 2006 - 04:12:00 EST.

Mailing list information is available at https://stat.ethz.ch/mailman/listinfo/r-help. Please read the posting guide before posting to the list.