Multiple Correspondence Analysis

[vtoledo]

Hi Dear Knimers,
I’m working on dimensionality reduction related to nominal variables and one technique used is the Multiple Correspondence Analysis

There is an implementation in R but I want to ensure that i’m using the platform at full capacity with already working nodes / workflows.

Do you know if there is any implementation already in Knime or anybody who is considering to do it?

Regards!
Victor

[lisovyi]

Welcome Victor,

there is a review of various dimentionality reduction techniques, that are available directly in KNIME: https://www.knime.com/blog/seven-techniques-for-data-dimensionality-reduction. However, for categorical variables several of them do not apply.

MSA in particular is not available directly in KNIME. However, one can easily use either R or python (i remember this implementation among others) integrations to run this technique in KNIME. I add @Maarit, as she might have further insights.

Cheers,
Misha

[Maarit]

Unfortunately I have nothing to add to Misha’s information. We’re currently updating the blog post about dimensionality reduction techniques. Thanks for this idea!