permute {seriation}

Permute the Order in Various Objects

Description

Provides the generic function and methods for permuting the order of various objects including vectors, lists, dendrograms (also hclust objects), the order of observations in a dist object, the rows and columns of a matrix or data.frame, and all dimensions of an array given a suitable ser_permutation object.

Usage

permute(x, order, ...)

## S3 method for class 'array'
permute(x, order, margin = NULL, ...)

## S3 method for class 'matrix'
permute(x, order, margin = NULL, ...)

## S3 method for class 'data.frame'
permute(x, order, margin = NULL, ...)

## S3 method for class 'table'
permute(x, order, margin = NULL, ...)

## S3 method for class 'numeric'
permute(x, order, ...)

## S3 method for class 'character'
permute(x, order, ...)

## S3 method for class 'list'
permute(x, order, ...)

## S3 method for class 'dist'
permute(x, order, ...)

## S3 method for class 'dendrogram'
permute(x, order, dist = NULL, ...)

## S3 method for class 'hclust'
permute(x, order, dist = NULL, ...)

Arguments

x

an object (a list, a vector, a dist object, a matrix, an array or any other object which provides dim and standard subsetting with "[").

order

an object of class ser_permutation which contains suitable permutation vectors for x. Alternatively, a character string with the name of a seriation method appropriate for x can be specified (see seriate()). This will perform seriation and permute x. The value TRUE will permute using the default seriation method.

...

if order is the name of a seriation method, then additional arguments are passed on to seriate().

margin

specifies the dimensions to be permuted as a vector with dimension indices. If NULL, order needs to contain a permutation for all dimensions. If a single margin is specified, then order can also contain a single permutation vector. margin are ignored.

dist

the distance matrix used to create the dendrogram. Only needed if order is the name of a seriation method.

Details

The permutation vectors in ser_permutation are suitable if the number of permutation vectors matches the number of dimensions of x and if the length of each permutation vector has the same length as the corresponding dimension of x.

For 1-dimensional/1-mode data (list, vector, dist), order can also be a single permutation vector of class ser_permutation_vector or data which can be automatically coerced to this class (e.g. a numeric vector).

For dendrogram and hclust, subtrees are rotated to represent the order best possible. If the order is not achieved perfectly then the user is warned. See also reorder.hclust() for reordering hclust objects.

Value

A permuted object of the same class as x.

Author(s)

Michael Hahsler

See Also

Other permutation: get_order(), permutation_vector2matrix(), ser_dist(), ser_permutation(), ser_permutation_vector()

Examples

# List data types for permute
methods("permute")

# Permute matrix
m <- matrix(rnorm(10), 5, 2, dimnames = list(1:5, LETTERS[1:2]))
m

# Permute rows and columns
o <- ser_permutation(5:1, 2:1)
o

permute(m, o)

## permute only columns
permute(m, o, margin = 2)

## permute using PCA seriation
permute(m, "PCA")

## permute only rows using PCA
permute(m, "PCA", margin = 1)

# Permute data.frames using heatmap seration (= hierarchical
#  clustering + optimal leaf ordering)
df <- as.data.frame(m)
permute(df, "Heatmap")

# Permute objects in a dist object
d <- dist(m)
d

permute(d, c(3, 2, 1, 4, 5))

permute(d, "Spectral")

# Permute a list
l <- list(a = 1:5, b = letters[1:3], c = 0)
l

permute(l, c(2, 3, 1))

# Permute to reorder dendrogram (see also reorder.hclust)
hc <- hclust(d)
plot(hc)

plot(permute(hc, 5:1))
plot(permute(hc, 5:1, incompartible = "stop"))

plot(permute(hc, "OLO", dist = d))
plot(permute(hc, "GW", dist = d))
plot(permute(hc, "MDS", dist = d))
plot(permute(hc, "TSP", dist = d))