KKNN vs flexclust?

[exsell-jc]

I would like to link this as context: https://logarithmic.net/pfh-files/blog/01618034037/k-means-better.html

In the tidymodels documentation for kknn, it cites this paper.

From a quick look, this may imply that the kknn does not apply an optimal vanilla KNN method. Disregarding other details now, it may be the case that the algorithm may need a minor adjustment to so called 'Ward + polished' linked above, where Ward minimises sum of squares, and complexity is O^2 instead of O^3 + some other polishing.

It could be that I'm missing one of the features in the function (possibly a hidden/undocumented in one of the arguments e.g. weight_func = 'optimal'). Please let me know if this is the case.

[EmilHvitfeldt]

Hello @exsell-jc 👋

Correct me if I'm wrong, but https://logarithmic.net/pfh-files/blog/01618034037/k-means-better.html is talking about K-Means the clustering algorithm, and https://parsnip.tidymodels.org/reference/details_nearest_neighbor_kknn.html talks about the K-Nearest Neighbors the regression/classification algorithm. I think you might have mixed the two (very similarly named) methods up 😄

[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.