Bug when trying to tune `hier_clust`

[davidrsch]

While trying to run a workflow in which in tune a hier_clust model I am receiving:

Error in check_grid() at tidyclust/R/tune_cluster.R:165:3:
! The provided grid has parameter column 'activation' that has not been marked for tuning by tune().

Which as the error states it's due to the fact that a parameter with identifier "activation" it's showing up in the grid, as you can see in the output of print(param_grid) of the following reprex:

install.packages(c("tidyclust", "dials", "parsnip"))
#> Installing packages into 'x/renv/library/windows/R-4.5/x86_64-w64-mingw32'
#> (as 'lib' is unspecified)
#> package 'tidyclust' successfully unpacked and MD5 sums checked
#> package 'dials' successfully unpacked and MD5 sums checked
#> package 'parsnip' successfully unpacked and MD5 sums checked
#> 
#> The downloaded binary packages are in
#>  C:\Users\x\AppData\Local\Temp\Rtmp2PKsV8\downloaded_packages

library(tidyclust)
library(dials)
#> Loading required package: scales
library(parsnip)
#> 
#> Attaching package: 'parsnip'
#> The following objects are masked from 'package:tidyclust':
#> 
#>     knit_engine_docs, list_md_problems

# Define the model with all tunable parameters
model <- hier_clust(
  num_clusters = tune(),
  linkage_method = tune()
) |>
  set_engine("stats") |>
  set_mode("partition")

# Define the parameter grid
param_grid <- dials::parameters(
  num_clusters(range = c(2, 3)),
  linkage_method(values = values_linkage_method)
)

# Show model and parameters
print(model)
#> Hierarchical Clustering Specification (partition)
#> 
#> Main Arguments:
#>   num_clusters = tune()
#>   linkage_method = tune()
#> 
#> Computational engine: stats
print(param_grid)
#> Collection of 2 parameters for tuning
#> 
#>    identifier         type    object
#>  num_clusters num_clusters nparam[+]
#>    activation   activation dparam[+]
#> 

I think this issue it's related to linkage_method due to that when I remove linkage_method tuning it works:

install.packages(c("tidyclust", "dials", "parsnip"))
#> Installing packages into 'x/renv/library/windows/R-4.5/x86_64-w64-mingw32'
#> (as 'lib' is unspecified)
#> package 'tidyclust' successfully unpacked and MD5 sums checked
#> package 'dials' successfully unpacked and MD5 sums checked
#> package 'parsnip' successfully unpacked and MD5 sums checked
#> 
#> The downloaded binary packages are in
#>  C:\Users\x\AppData\Local\Temp\Rtmp2PKsV8\downloaded_packages

library(tidyclust)
library(dials)
#> Loading required package: scales
library(parsnip)
#> 
#> Attaching package: 'parsnip'
#> The following objects are masked from 'package:tidyclust':
#> 
#>     knit_engine_docs, list_md_problems

# Define the model with all tunable parameters
model <- hier_clust(
  num_clusters = tune() # ,
  # linkage_method = tune()
) |>
  set_engine("stats") |>
  set_mode("partition")

# Define the parameter grid
param_grid <- dials::parameters(
  num_clusters(range = c(2, 3)) # ,
  # linkage_method(values = values_linkage_method)
)

# Show model and parameters
print(model)
#> Hierarchical Clustering Specification (partition)
#> 
#> Main Arguments:
#>   num_clusters = tune()
#>   linkage_method = complete
#> 
#> Computational engine: stats
print(param_grid)
#> Collection of 1 parameters for tuning
#> 
#>    identifier         type    object
#>  num_clusters num_clusters nparam[+]
#> 

[lgaborini]

If you name the linkage_method argument, it works:

# Define the parameter grid
param_grid <- dials::parameters(
  num_clusters(range = c(2, 3)),
  linkage_method = linkage_method(values = values_linkage_method)
)