Create a grid of `mdist()` method specifications

Construct a tibble of valid method specifications that can be passed to [benchmark_mdist()] or iterated over manually to benchmark [mdist()] across multiple configurations.

Usage

all_dist_method_specs(
  method_cat = NULL,
  method_num = NULL,
  preset = NULL,
  commensurable = NULL
)

Arguments

method_cat

Optional character vector restricting the categorical methods used in custom specifications.

method_num

Optional character vector restricting the numerical preprocessing methods used in custom specifications.

preset

`NULL` (the default) returns custom combinations only. A character vector selects named presets; `"all"` selects all named presets. Include `"custom"` explicitly to also return custom combinations, for example `c("all", "custom")`. Component filters apply only to custom rows.

commensurable

Optional logical vector restricting the commensurable values used in custom specifications.

Value

A tibble where each row represents one valid `mdist()` specification. The tibble contains `preset`, `method_cat`, `method_num`, and `commensurable`.

Details

Select named presets or construct custom specifications from the categorical methods and numerical preprocessing options listed in [dist_methods_tbl()].

Rows with `preset = "custom"` use the explicit categorical, numerical, and commensurability options. Named presets determine those settings internally.

With `commensurable = TRUE`, custom grids retain only `"std"` and `"pc_scores"` numerical preprocessing. Variable-wise affine scaling cancels under commensurability, so `"robust"` is omitted as redundant with `"std"`. The existing exclusions of `"none"` and `"range"` are retained. Range scaling clips new observations outside the training range, so its equivalence on training data does not extend to all new observations. These restrictions affect grid construction, not the options accepted by [mdist()].

Named preset rows retain `NA` in their component columns: their settings are determined by the preset. Supplying component filters without requesting custom rows gives an error rather than silently ignoring those filters.

Examples

all_dist_method_specs()
#> # A tibble: 413 × 4
#>    preset method_cat method_num commensurable
#>    <chr>  <chr>      <chr>      <lgl>        
#>  1 custom HL         none       FALSE        
#>  2 custom HL         pc_scores  FALSE        
#>  3 custom HL         pc_scores  TRUE         
#>  4 custom HL         range      FALSE        
#>  5 custom HL         robust     FALSE        
#>  6 custom HL         std        FALSE        
#>  7 custom HL         std        TRUE         
#>  8 custom HLeucl     none       FALSE        
#>  9 custom HLeucl     pc_scores  FALSE        
#> 10 custom HLeucl     pc_scores  TRUE         
#> # ℹ 403 more rows
all_dist_method_specs(preset = c("gower", "u_indep", "u_dep"))
#> # A tibble: 3 × 4
#>   preset  method_cat method_num commensurable
#>   <chr>   <chr>      <chr>      <lgl>        
#> 1 gower   NA         NA         NA           
#> 2 u_dep   NA         NA         NA           
#> 3 u_indep NA         NA         NA           
all_dist_method_specs(preset = "all")
#> # A tibble: 10 × 4
#>    preset    method_cat method_num commensurable
#>    <chr>     <chr>      <chr>      <lgl>        
#>  1 euclidean NA         NA         NA           
#>  2 gower     NA         NA         NA           
#>  3 hl        NA         NA         NA           
#>  4 u_dep     NA         NA         NA           
#>  5 u_dep_bw  NA         NA         NA           
#>  6 u_indep   NA         NA         NA           
#>  7 u_mix     NA         NA         NA           
#>  8 dkss      NA         NA         NA           
#>  9 gudmm     NA         NA         NA           
#> 10 mod_gower NA         NA         NA           
all_dist_method_specs(preset = c("all", "custom"))
#> # A tibble: 423 × 4
#>    preset    method_cat method_num commensurable
#>    <chr>     <chr>      <chr>      <lgl>        
#>  1 euclidean NA         NA         NA           
#>  2 gower     NA         NA         NA           
#>  3 hl        NA         NA         NA           
#>  4 u_dep     NA         NA         NA           
#>  5 u_dep_bw  NA         NA         NA           
#>  6 u_indep   NA         NA         NA           
#>  7 u_mix     NA         NA         NA           
#>  8 dkss      NA         NA         NA           
#>  9 gudmm     NA         NA         NA           
#> 10 mod_gower NA         NA         NA           
#> # ℹ 413 more rows
all_dist_method_specs(method_cat = c("tvd", "le_and_ho"))
#> # A tibble: 14 × 4
#>    preset method_cat method_num commensurable
#>    <chr>  <chr>      <chr>      <lgl>        
#>  1 custom le_and_ho  none       FALSE        
#>  2 custom le_and_ho  pc_scores  FALSE        
#>  3 custom le_and_ho  pc_scores  TRUE         
#>  4 custom le_and_ho  range      FALSE        
#>  5 custom le_and_ho  robust     FALSE        
#>  6 custom le_and_ho  std        FALSE        
#>  7 custom le_and_ho  std        TRUE         
#>  8 custom tvd        none       FALSE        
#>  9 custom tvd        pc_scores  FALSE        
#> 10 custom tvd        pc_scores  TRUE         
#> 11 custom tvd        range      FALSE        
#> 12 custom tvd        robust     FALSE        
#> 13 custom tvd        std        FALSE        
#> 14 custom tvd        std        TRUE