search

tidymodels / embed

Extra recipes for predictor embeddings
https://embed.tidymodels.org
Other
142 stars 19 forks source link

Feature request: step_mca() or multiple correspondence analysis -- dimension reduction for categorical variables #221

Open exsell-jc opened 2 years ago

exsell-jc commented 2 years ago

Similar to step_pca() or step_umap() where it may be useful for dimensional reduction, step_mca() may seem useful for reducing dimensions for categorical predictors.

juliasilge commented 2 years ago

Thanks for this idea @exsell-jc! Do you typically use MASS::mca() or something else for this kind of task?

exsell-jc commented 2 years ago

Thanks for this idea @exsell-jc! Do you typically use MASS::mca() or something else for this kind of task?

Sure thing @juliasilge

Actually, I'm unsure if MASS is better or FactoMineR. I can't really seem to find any benchmarks or comparisons. MASS does seem to be downloaded twice as often and more actively updated when looking at these two links:

https://www.rdocumentation.org/packages/FactoMineR/versions/2.4 https://www.rdocumentation.org/packages/MASS/versions/7.3-56