Skip to contents

Multidimensional Scaling

Source: R/mv_stats.R
mv_mds.Rd

Computes a multidimensional scaling configuration from the distances of the selected numeric columns: classical (metric) MDS via stats::cmdscale or non-metric MDS via MASS::isoMDS.

Usage

mv_mds(
  data,
  cols,
  k = 2,
  metric = TRUE,
  dist_method = "euclidean",
  scale = TRUE,
  maxit = 50
)

Arguments

data

A data frame.

cols

<tidy-select> Numeric columns to scale.

k

Number of dimensions.

metric

Logical. If TRUE (default), classical metric MDS (cmdscale); if FALSE, non-metric MDS (MASS::isoMDS).

dist_method

Distance method passed to stats::dist.

scale

Logical. If TRUE (default), columns are standardized first.

maxit

Maximum iterations for MASS::isoMDS.

Value

An object of class mv_mds: a list with components points (tibble with .row and Dim1...), eigen (tibble with dimension and eigenvalue, metric MDS only), stress (single-row tibble, non-metric MDS only), gof (single-row tibble with gof1, gof2, metric MDS only), and meta.

Examples

r = mv_mds(mtcars, cyl:carb, k = 2)
r$gof
#> # A tibble: 1 × 2
#>    gof1  gof2
#>   <dbl> <dbl>
#> 1 0.841 0.841
plot_mv_mds(r)