Alternatives for the PCA Node / Determine the ideal number of clusters

[gnime]

As this thread states: Knime PCA node convergence failed, the PCA Node is not working correctly.

I am trying to replicate this worklfow with my own data: https://www.knime.com/blog/topic-extraction-optimizing-the-number-of-topics-with-the-elbow-method

Is there an alternative to the PCA node that I can use in my workflow instead ?

Or in general are there other example workflows to determine the ideal number of clusters/topics for the LDA?

[ScottF]

Hi @gnime -

As a temporary workaround, you can try using the deprecated version of the PCA node. Ours devs have been working on a fix for PCA - it’s currently in verification. Sorry for the trouble.

[gnime]

Hi @ScottF ,

Thanks for your answer.
I am already using the deprecated version, but sometimes it does not work properly.

Are there maybe any other example workflows which let you determine the optimal topics for the LDA?

[ipazin]

@gnime

As this thread states: Knime PCA node convergence failed , the PCA Node is not working correctly.

Fixed with 4.1.0. version.

Br,
Ivan