Is it possible to get probabilities of class membership based on Multiple Correspondence Analysis which uses hierarchical clustering?

I am thinking along the lines of Latent Class Analysis that provides these posterior probabilities.

I am guessing the answer is no.

Secondly, how can I assess model fit from the results of MCA?

Thanks in advance.

  • $\begingroup$ Can you tell us what is your ultimate goal, and something about your data? $\endgroup$ Nov 16 '20 at 17:59
  • 1
    $\begingroup$ Thanks Kjetil for taking the time to respond. The data set consists of approx. 1900 rows (1 row per participant) and 14 columns. The 14 columns are all categorical variables and represent exposures to various pollutants. I have applied MCA via library(FactoMineR) to group participants according to exposure to the pollutants. (BTW, FactoMineR is a very useful package for MCA). To compare results, I have also done a latent class analysis with the same data. WIth LCA, I get probabilities of belonging to the latent classes. I don't seem to get that with MCA which uses hierarchical clustering. $\endgroup$ Nov 21 '20 at 14:49
  • $\begingroup$ You need to give still more details, and please as an edit to the post and not only as comments. What is LCA? Hierarchcal clustering is not a part of MCA as I have learnt it ... What are your 14 variables? MCA uses usually count data, but exposure does not sound like a count, or is it 0/1? ... $\endgroup$ Nov 21 '20 at 19:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.