Compare LOVO diagnostics across multiple distance specifications

This function compares **leave-one-variable-out (LOVO)** diagnostics across multiple distance definitions supported by manydist.

Usage

compare_lovo_mdist(
  x,
  methods,
  mds = FALSE,
  dims = 2,
  keep_dist = FALSE,
  .progress = FALSE,
  ...
)

Arguments

x

A data frame or tibble containing the predictors.

methods

A **named list** describing the distance specifications to compare. Each element must be a list of arguments passed to lovo_mdist.

For example:

methods = list(
  gower = list(preset = "gower"),
  u_dep = list(preset = "unbiased_dependent"),
  custom = list(
    preset = "custom",
    method_cat = "matching",
    method_num = "std",
    commensurable = TRUE
  )
)

mds

Logical. If `TRUE`, compute MDS-based congruence and alienation diagnostics for every distance specification. The default is `FALSE`.

dims

Number of dimensions used for the MDS configuration when `mds = TRUE`. Ignored when `mds = FALSE`.

keep_dist

Logical; if TRUE, distance matrices from the LOVO computations are retained. This increases memory usage.

.progress

Logical; if TRUE, progress messages are printed while computing LOVO diagnostics for each method.

...

Additional arguments passed to lovo_mdist unless overridden in the method-specific argument list.

These may include optional clustering diagnostics, for example cluster_k, cluster_methods, and hclust_method.

Value

An object of class MDistLOVOCompare containing:

results

A tibble with one row per method-variable combination.

methods

The list of distance specifications used.

mds

Whether MDS diagnostics were computed.

dims

Number of MDS dimensions used, or `NULL` when `mds = FALSE`.

n_obs

Number of observations in the dataset.

Details

For each distance specification, the function:

1. Computes the full mixed-type distance using mdist().

2. Recomputes the distance repeatedly leaving out one variable at a time.

3. Measures the impact of each variable using mean absolute deviation (MAD) and, when requested, MDS- or clustering-based diagnostics.

The results are combined across methods and returned as an MDistLOVOCompare object, which supports print(), summary(), and ggplot2::autoplot().

In autoplot(), reorder = TRUE ranks variables by the selected metric averaged across methods, separately within each variable type. Categorical variables precede numerical variables, preserving the blue/red background bands. top_n still selects variables globally before ordering. If type information is absent, ordering is global.

See also

lovo_mdist, mdist

Examples

if (FALSE) { # \dontrun{
library(manydist)
library(palmerpenguins)

data <- penguins |>
  dplyr::select(-species) |>
  tidyr::drop_na()

cmp <- compare_lovo_mdist(
  x = data,
  methods = list(
    gower = list(preset = "gower"),
    u_dep = list(preset = "unbiased_dependent")
  ),
  cluster_k = 3
)

summary(cmp)
autoplot(cmp, metric = "mad_importance")
autoplot(cmp, metric = "pam_importance")
} # }