search

juliasilge / juliasilge.com

My blog, built with blogdown and Hugo :link:
https://juliasilge.com/
40 stars 27 forks source link

Explore art media over time in the #TidyTuesday Tate collection dataset | Julia Silge #15

Open utterances-bot opened 3 years ago

utterances-bot commented 3 years ago

Explore art media over time in the #TidyTuesday Tate collection dataset | Julia Silge

Check residuals and other model diagnostics for regression models trained on text features, all with tidymodels functions.

https://juliasilge.com/blog/tate-collection/

mwilson19 commented 3 years ago

pull_workflow_fit(art_final$.workflow[[1]])

That's a strange syntax, funny I was having issues with glmnet in tidymodels as I was misunderstanding how to pull the right iteration with the specified penalty, interesting to see how you do this. I think the default from vip package is to take the last iteration as glmnet runs many iterations of lambda (penalty) by default too. Anyways thanks for this as always Julia!

conlelevn commented 2 years ago

Nice work Julia as always, sparse data is a new knowledge that I have not usually used before, if we did not apply this sparse data into our model, would it be any bad effect on the result? thanks

juliasilge commented 2 years ago

@conlelevn The main benefit of using a sparse representation is a speed-up in training in some situations. You can see an example of this here.