|
|
Probabilistic latent semantic analysis |
|
Message/Author |
|
|
Greetings, I am just getting started with Mplus. Coincidentally, I'm investigating the behaviour of so-called probabilistic latent semantic analysis (Hoffman 1999) http://www.cs.brown.edu/~th/papers/Hofmann-UAI99.pdf , which is a fancy name for a latent mixture model with nominal observed and unobserved vars. Parameters are estimated using EM. I have implemented the procedure in software, but I am wondering if Mplus is capable of fitting these models as well -- and if so, whether there are scalability issues with using Mplus that make in impractical for this application. Best, jds |
|
|
A model with nominal observed and categorical variables using EM is possible in Mplus. See Example 7.7 in the user's guide. |
|
Back to top |
|
|