rename extract_cluster and extract_cluster_assignments

[kbodwin]

As I work with these, I don't like "extract" as the verb - especially because we might be considering extraction style clustering methods in the future...

Being nitpicky, I also thing extract_cluster should be wordier. (Is the "cluster" the centroids? The observations?).

For the sake of starting conversation, I'll propose:

get_cluster_centers - with future optional arguments for different notions of centers beyond geometric means

I'm also still on the fence as to whether extract_cluster_assignments needs to exist at all. In tidymodels, the training data is labelled using predict. Since we want predict to always return the original cluster assignments, it may be all we need. Although the dedicated function does avoid duplicate computation of cluster assignments, I suppose...

[EmilHvitfeldt]

I picked the extract_ verb because we already have a number of extract_*() functions in the tidymodels. extract_mold(), extract_fit_parsnip() etc, etc. I would be okay switching if we can find a more appropriate verb.

Being nitpicky, I also thing extract_cluster should be wordier. (Is the "cluster" the centroids? The observations?).

Agree! right now it returns the centroids.

For now I like to keep extract_cluster_assignments() around because it removed the need for duplicate calculations. Especially since the model objects have that information embedded in them. It seems weird not to have a way to extract it.
That being said, the result of extract_cluster_assignments(model) and predict(model, training_data) should always be the same.

[kbodwin]

Hmmm, that's a good argument for keeping extract_, but then we'd need a new word for when you identify a single cluster on each run of the algorithm. I'll mull this over.

Re: cluster assignments - I definitely take your point, and I know I've advocated for an easier way to extract info from supervised models too, so I guess this falls under that umbrella.

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.