Distance matrix with inter feature(variable) distances based on PCA

Computes distance between features(variables) weighing the loadings proportional to corresponding eigenvalues

dist_pca(model, ...)

Arguments

model

Object of class 'prcomp' or 'princomp'.
...

Arguments to be passed to 'proxy::dist'

Details

See the book: 'An introduction to applied multivariate analysis with R' by Everitt (et al) section 3.6 on 'Rescaling the principal components'

Examples

pca_iris          <- stats::prcomp(scale(iris[,1:4]))
feature_dist_iris <- dist_pca(pca_iris)
feature_dist_iris
#>              Sepal.Length Sepal.Width Petal.Length
#> Sepal.Width    2.36441184                         
#> Petal.Length   0.38799175  2.61263496             
#> Petal.Width    0.36928628  2.55802891   0.09765659
plot(stats::hclust(feature_dist_iris))

pca_iris2         <- stats::princomp(covmat = cor(iris[,1:4]), cor = TRUE)
feature_dist_iris2 <- dist_pca(pca_iris2)
feature_dist_iris2
#>              Sepal.Length Sepal.Width Petal.Length
#> Sepal.Width    2.36441184                         
#> Petal.Length   0.38799175  2.61263496             
#> Petal.Width    0.36928628  2.55802891   0.09765659
all.equal(feature_dist_iris, feature_dist_iris2)
#> [1] TRUE
plot(stats::hclust(feature_dist_iris))