topic Re: How does Latent Class Analysis work on Text Explorer in Discussions

How does Latent Class Analysis work on Text Explorer

statlover — Fri, 22 Nov 2019 20:44:57 GMT

On the Multivariate Analysis Tool kit Latent Class Analysis is used to predict clusters on Categorical data (Example Health Risk Survey on JMP Library). How is that principle used on Text Data? We get a Document Term Matrix which is either binary (numeric) , Frequency or TF-IDF. We do not get a DTM of Categorical data. So how does LCA work in this case? Does the binary DTM get converted internally to categorical DTM?

Re: How does Latent Class Analysis work on Text Explorer

Mark_Zwald — Fri, 22 Nov 2019 21:02:01 GMT

You can learn more about LCA within Text Explorer here: https://www.jmp.com/support/help/en/15.0/#page/jmp/latent-class-analysis.shtml

And more about the LCA platform here: https://www.jmp.com/support/help/en/15.0/#page/jmp/latent-class-model-fit.shtml#ww383604

Refer to the note at the bottom: The LCA algorithm that is used in the Text Explorer platform takes advantage of the specific structure of the document term matrix. For this reason, the LCA results in the Text Explorer platform do not exactly match the results in the Latent Class Analysis platform.

Re: How does Latent Class Analysis work on Text Explorer

Mark_Bailey — Sat, 23 Nov 2019 16:41:04 GMT

The LCA is an unsupervised learning method in Text Explorer. It discovers clusters of documents. It is not a classifier.

Re: How does Latent Class Analysis work on Text Explorer

statlover — Mon, 25 Nov 2019 18:28:35 GMT

Thank you very much. I tried to reproduce the results on the Multivariate LCA platform after converting the DTM matrix to a categorical matrix. Since the output is very different, I could not come to any conclusion. But your feedback is valuable.

Re: How does Latent Class Analysis work on Text Explorer

statlover — Mon, 25 Nov 2019 18:31:38 GMT

Thanks Mark Bailey. I understand it is unsupervised learning algorithm. I was trying to see if I could get the same result on the Multivariate platform and the Text Explorer platform. It appears the mechanism is different. Thanks for responding to my question.