PCA and UMAP with tidymodels and cocktail recipes

[@AlexBorgesBBorg]

Im a beginnerish and boy do I wish I had found your videos earlyer. I bow in signal of respect.

[@stoianandreimircea1509]

Thank you again Julia. Bravo on the speed and advance topics covered.

[@brettjuelz2373]

Instablaster

[@albertcampillo]

Amazing Julia! You make PCA & UMAP concepts crystal clear!

[@renanxcortes2]

Great stuff!! Watching your programming fluency (specially in the ggplot2 part) was absolutely amazing! Thanks for posting the video!

[@gavinaustin4474]

Really enjoying these videos, Julia. Please keep 'em coming. I'm planning a quantitative corpus-based study of the use of metaphors in classical music reviews, and I'm considering how I should approach it. This is a new area for me.

[@Madeinoz1967]

Awesome stuff, I love your style in presenting the topic, and has really helped with tidymodels and PCA. Many Thanks

[@danilomlustosa]

Thanks Julia, that was amazing!

[@hesamseraj]

Bravo, bravo, bravo I am supposed to write my final papers but instead, I am trying to learn from you.

[@erickknackstedt3131]

Amazing! Great stuff. Thank you.

[@pritomroy2465]

Wow, umap is already in r ! just moved to umap over tsne.

[@williamhenry7226]

Great presentation! Dashboard issues. Power BI and umap are not dancing together.

[@konsmberg]

Love your videos, thank you. Hope to see at some point PCA that does NOT use recipes.

[@JuliaSilge]

If you don't want to use tidymodels for PCA, you might like Claus Wilke's approach here: clauswilke.com/blog/2020/09/07/pca-tidyverse-style/

[@konsmberg]

@@JuliaSilge thank you!!!

[@wapsyed]

Your videos are therapeutic haha

[@christospapadopoulos7704]

Julia you are a goddess!! Thank you.

[@nicolacatena9025]

Julia you are amazing! :)

[@dariyatukhmetova1172]

Great tutorial, thank you

[@ytoren]

Would it not make sense to use REGEX one-liners to generalise a lot of the replacements? match = "(orange|grapefruit|lemon|...)\s?juice", replace = "$1" match = " ?([0-9])/([0-9])", replace = "$1 $2" %>% str_split(" ") %>% function(x) {x[1]/x[2]}

[@JuliaSilge]

Yes, I think I even said that at 6:54 or so, that it would have been better to use a little function.

[@Levy957]

i love your videos

[@elysium55916]

Thanks Julia! Can you ggplot a rotating 3D plot with the first 3 components?

[@JuliaSilge]

You might check out the tourr package for something like that: ggobi.github.io/tourr/

[@patolobos8266]

Thanks Julia! :-)

[@Actanonverba01]

Great Presentation. However I would have liked a little more explanation on the UMAP, But otherwise 9.5/10.

[@JOHNSMITH-ve3rq]

Julia - considered putting all this code on github? Would make it a lot easier to reference! Thank you.

[@JuliaSilge]

You can find the code that builds my blog here: github.com/juliasilge/juliasilge.com The blog posts are in /content/blog/

[@PA_hunter]

Hi Julia, I am working on a project and am stuck between two general types of errors when prepping my PCA recipe: the first is this: Error in svd(x, nu = 0, nv = k) : infinite or missing values in 'x' I interpret this to be some issue with missing data so I have tried both step_naomit and step_other as solutions. When I go to prep these updated recipes, I get errors along the lines of: ℹ It must be numeric or character. Backtrace: 1. recipes::prep(pca_rec, training = datatest) 26. rlang::cnd_signal(``) Do you know what might be going wrong? Thanks!

[@JuliaSilge]

Hmmmm, it's hard to say from just this information. I suggest that you create a reprex demonstrating your problem: www.tidyverse.org/help/ And post on RStudio Community: rstd.io/tidymodels-community That is a good place to get help for R coding problems. Good luck!

[@teorems]

Bitters are good for health!

[@stuartedwards3622]

Love your work Julia, but could you please ask KZbin to stop subtitling you as Julia Silly as you clearly are not 😂

[@JuliaSilge]

HA I really should work on doing my own subtitles instead of letting KZbin autogenerate them!!