Initialization PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
 Joshua Rosenberg posted on Monday, January 01, 2018 - 4:53 pm
Hi, thank you for developing and supporting such a helpful tool. I have used the mclust software to estimate models like those estimated as part of Latent Profile Analysis, and am curious about whether initializing the E-M algorithm with hierarchical clustering rather than random starts seems like a reasonable idea. I know that I can implement this by providing start values from a hierarchical clustering (in Mplus), but was curious about its features in relation to random starts in the context of carrying out mixture modeling or Latent Profile Analysis

Thank you!
 Bengt O. Muthen posted on Tuesday, January 02, 2018 - 2:04 pm
I would think using random starts works better and is at least as good, particularly since you can use so many of them.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message