search

tidymodels / tidyclust

A tidy unified interface to clustering models
https://tidyclust.tidymodels.org/
Other
105 stars 14 forks source link

Error message when nstart is misspecified #181

Open katieburak opened 3 months ago

katieburak commented 3 months ago

When performing k-means, if you use the wrong name for nstart (e.g., nstarts) it does not throw you an error. For example, the following code chunk runs without error:

kmeans_spec_tune <- k_means(num_clusters = tune()) |> set_engine("stats", nstarts = 10) kmeans_spec_tune

However, when you add this to a workflow you get the following error:

`x Apparent: preprocessor 1/1, model 1/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 2/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 3/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 4/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 5/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 6/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 7/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 8/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 9/10: Error in stats::kmeans(data, centers, ...)...

x Apparent: preprocessor 1/1, model 10/10: Error in stats::kmeans(data, centers, ......

Warning message: “All models failed. See the .notes column.” Error in estimate_tune_results(): ! All models failed. Run show_notes(.Last.tune.result) for more information. Traceback:

  1. collect_metrics(tune_cluster(add_model(add_recipe(workflow(), . kmeans_recipe), kmeans_spec_tune), resamples = apparent(clean_beer), . grid = beer_ks))
  2. collect_metrics.tune_results(tune_cluster(add_model(add_recipe(workflow(), . kmeans_recipe), kmeans_spec_tune), resamples = apparent(clean_beer), . grid = beer_ks))
  3. estimate_tune_results(x)
  4. rlang::abort("All models failed. Run show_notes(.Last.tune.result) for more information.")
  5. signal_abort(cnd, .file)`

It would be helpful and informative for the error to come earlier so it is easier to diagnose the problem. Thank you!

EmilHvitfeldt commented 3 months ago

Hello @katieburak 👋

Thank you for the bug report! I'll try to handle next time I get to do some tidyclust work