EM (Expectation Maximization) Algorithm

Hello:

I just wanted to know whether EM (Expectation Maximization) algorithm is available in GAUSS.

Thank you very much.

Annesha

5 Answers



0



The "Expectation/Maximization" algorithm is different for each type of problem so its not possible to write a general program for it.   GAUSS is a full-featured programming language so it would be straightforward to write code for the expectation part and then use any of the optimization programs in GAUSS to produce the maximization part.



0



Without knowing further information about the desired application of the EM algorithm, it is difficult to give further insight into the tools available in GAUSS to assist with EM.

However, there are examples of third party code that use EM. For example Dr. James Hamilton at UCSD provides GAUSS EM code for estimating Markov Switching models. This can be found at http://weber.ucsd.edu/~jhamilto/software.htm

As noted on the website "These are written in the GAUSS programming language. They do not require use of the GAUSS numerical optimization procedures and should work with little or no change on any version of GAUSS..."

Eric

90


0



Hello: Thank you for your email. I need to code a latent class model using EM algorithm.

Thanks.

Annesha



0



Though there is no internal program for estimating latent class models using the EM algorithm, it may be helpful to look at the paper and GAUSS code provided by Caroline Beunckens, Geert Molenberghs, Geert Verbeke, and Craig Mallinckrodt at

http://www.biometrics.tibs.org/datasets/060708R2_manual_describing_computer_code.pdf.

I cannot provide direct verification or feedback regarding the code. However, you might be able to get further information by directly contacted one of the authors.

 

Eric

90


0



Hi Annesha:

Did you find any gauss code to model the latent class model using the EM algorithm or any reference for the same? I am also looking for the code.

Thanks and Regards,

Sangram Nirmale

 

Your Answer

5 Answers

0

The "Expectation/Maximization" algorithm is different for each type of problem so its not possible to write a general program for it.   GAUSS is a full-featured programming language so it would be straightforward to write code for the expectation part and then use any of the optimization programs in GAUSS to produce the maximization part.

0

Without knowing further information about the desired application of the EM algorithm, it is difficult to give further insight into the tools available in GAUSS to assist with EM.

However, there are examples of third party code that use EM. For example Dr. James Hamilton at UCSD provides GAUSS EM code for estimating Markov Switching models. This can be found at http://weber.ucsd.edu/~jhamilto/software.htm

As noted on the website "These are written in the GAUSS programming language. They do not require use of the GAUSS numerical optimization procedures and should work with little or no change on any version of GAUSS..."

0

Hello: Thank you for your email. I need to code a latent class model using EM algorithm.

Thanks.

Annesha

0

Though there is no internal program for estimating latent class models using the EM algorithm, it may be helpful to look at the paper and GAUSS code provided by Caroline Beunckens, Geert Molenberghs, Geert Verbeke, and Craig Mallinckrodt at

http://www.biometrics.tibs.org/datasets/060708R2_manual_describing_computer_code.pdf.

I cannot provide direct verification or feedback regarding the code. However, you might be able to get further information by directly contacted one of the authors.

 

0

Hi Annesha:

Did you find any gauss code to model the latent class model using the EM algorithm or any reference for the same? I am also looking for the code.

Thanks and Regards,

Sangram Nirmale

 


You must login to post answers.

Have a Specific Question?

Get a real answer from a real person

Need Support?

Get help from our friendly experts.

Try GAUSS for 14 days for FREE

See what GAUSS can do for your data

© Aptech Systems, Inc. All rights reserved.

Privacy Policy