0

I am interested in the Gaussian mixture model. I read about it and I think I am good with it. However, found that there is something called truncated Gaussian mixture model, which I do not understand. I read some papers, but still do not understand the meaning of it! Does the truncated gaussian mixture model mean that we do not use all the observations in the data?! What is the importance of it? What is the benefit of using it over the Gaussian mixture model?

Here is one reference:

McLachlan & Jones 1988

and here is the other one

Lee & Scott 2010

and a new one is here:

https://www.tandfonline.com/doi/abs/10.1080/00949659708811866

Maryam
  • 1,012
  • 1
  • 5
  • 20
  • 1
    It may just be a mixture of [truncated normal distributions](https://en.wikipedia.org/wiki/Truncated_normal_distribution), which sounds a reasonable model in cases where you would expect truncation. We have a [tag:truncated-normal-distribution] tag. Can you give a little more context, perhaps some of the papers where you found the concept? – Stephan Kolassa Jul 09 '21 at 14:05
  • 1
    Could you provide references to the papers? It would be easier to answer knowing the context. – Tim Jul 09 '21 at 14:45
  • 1
    @StephanKolassa Thanks a lot for your comment. I have added some references. – Maryam Jul 09 '21 at 16:10
  • 1
    @Tim Thanks a lot for your comment. I have included some references. – Maryam Jul 09 '21 at 16:10
  • These are *mixture models* for fitting "truncated data." That's different than fitting mixtures of truncated Gaussians to data! You should therefore research the meaning of truncation in statistics. See https://stats.stackexchange.com/questions/144041. – whuber Jul 09 '21 at 16:37

0 Answers0