Spice Girls was my fav band as a queer little kid. I'm in grad school learning NLP and this video is a true full-circle moment for me! Thanks so much for sharing.
Great video as always. Is there a tidymodels implementation of stm, or another easy way to do CV for the number of topics? Recently I'm really loving how easy it is to customize tuning and resampling with tidymodels.
No, the focus of tidymodels is largely predictive or supervised algorithms (we do have support for unsupervised algorithms in recipes, especially those that can be used as preprocessing for supervised ML). I have a blog post here that shows an approach for training many different topic models and choosing K: juliasilge.com/blog/evaluating-stm/
I select all the code in the doc with Cmd+A and then I reindent the lines with Cmd+1 (I'm on a Mac). That is one of my most used keyboard shortcuts in RStudio! You can see the shortcuts in the IDE under Tools, or they are here: support.rstudio.com/hc/en-us/articles/200711853-Keyboard-Shortcuts-in-the-RStudio-IDE
Within estimateEffect, you have 1:4 as the first arg. Is that referencing "topics 1 through 4"? As always, thank you for sharing your time and expertise.
