Dimensionality reduction of United Nations voting patterns

[@91chaves]

It was funny so spot Yugoslavia and Czechoslovakia there :D Pretty cool project Julia, keep them coming. I learn a lot from your videos.

[@HaiLeQuang]

@julia, pls make more video. Your channel is my valuable resource for learning

[@mikegaunt837]

Quick question: how do you reconcile applying PCA reduction to discrete data like in this case (or for word counts/one hot/etc)? I thought that if your variables don't belong on a coordinate plane, then do not apply PCA to them. I've been struggling with this concept especially with unsupervised learning with text data and sparse data. Are there any plans to bring in other unsupervised methods into recipes a la clusterR, cluster, factormineR? If not, are there any methods that you like to use yourself when doing EDA/unsupervised leaning?

[@hesamseraj]

Amazing, Thank you very much. I eagerly wait for your screencast

[@aruncps]

interesting visualization 🥰 . Thank you for sharing.

[@nagytamashungary]

Great video, thanks! I was wondering if it is possible to handle the number of components as a hyperparameter, so as to find an optimal solution by tuning?

[@JuliaSilge]

Yes! Look at this section for how to do that: www.tmwr.org/grid-search.html#evaluating-grid

[@davidjackson7675]

Very interesting, Thanks for sharing.

[@cheninitayeb6590]

thank you lot please Julia if you could please session with bioconductor ,thank you so much

[@davidjackson7675]

It could be that some countries were not part of UN in the earlier years..

[@Mohamed-sq8od]

hey , love your videos and thanks to you i started working with tidymodels ..... i'm working on a project based on string similarities , i was wondering can the same thing be done after i tokenize the strings ??

[@JuliaSilge]

Yes, you can implement many unsupervised ML approaches on word counts or tf-idf. You could do PCA, or you could find clusters, like I demonstrate here: stackoverflow.com/questions/66030942/tidytext-clustering/66056334#66056334

[@pwylll]

Excellent content as always! What RStudio theme are you using?

[@JuliaSilge]

It's one of the ones available in rsthemes, Oceanic Plus: github.com/gadenbuie/rsthemes

[@renatorsprojeto3d643]

Hello Julia, how are you? have you managed to integrate the "modelplotr" package with tidymodels? I tried and was unsuccessful. My goal is to create a plot cumulative gains cumulative lift response cumulative response costs & revenues profit return on investment using templates that I created on tidymodels, but only works with templates created by the "caret / mlr" packages

[@JuliaSilge]

If you are a user of modelplotr, I would suggest opening an issue about supporting tidymodels workflows, much like they support caret and mlr: github.com/jurrr/modelplotr/issues We are happy to help out or give advice. In the meantime, you might have luck by accessing the underlying model fits from a parsnip model via `$fit` and using that with modelplotr.

[@renatorsprojeto3d643]

@@JuliaSilge thank you so much!

[@caiogracoroza6033]

PC4 = Arms control and Nuclear weapons, not human rights. Makes even more sense.