validate: Partition Quality Assessment for '"VegsoupPartition"' Objects

Description Usage Arguments Details Value Note Author(s) See Also Examples

Description

Partition analysis (Partana) evaluates the within-cluster to among-cluster similarity as a measure of cluster validity. Silhouette information (Silhouette) or Dissimilarity Diameters (Disdiam) can also be used to evaluate a partitioning of an object. All three are distance based methods.

Table Deviance (Tabdev) assess the clarity of the partitioning by permuting the columns (species) of the community matrix (original data) and calculates the probability of observing as low a deviance as observed.

All methods are interfaces to corresponding functions in package optpart.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
## S4 method for signature 'VegsoupPartition'
silhouette(x, ...)

## S4 method for signature 'VegsoupPartition'
partana(x, verbose = FALSE, ...)

## S4 method for signature 'VegsoupPartition'
disdiam(x, ...)

## S4 method for signature 'VegsoupPartition'
tabdev(x, numitr = 99, verbose = FALSE, ...)

Arguments

x

A "VegsoupPartition" object.

numitr

Number of iterations to perform in calculating the probability of obtaining as effective a classification as observed.

verbose

additional console output.

...

Arguments passed to as.dist (Partana, Silhouette and Disdiam) or link{as.matrix} (Tabdev).

Details

Partana (partition analysis) evaluates the within-cluster to among-cluster similarity of partitions as a measure of cluster validity. See partana for details. disdiam is also a cluster validation routine.

As opposed to partition analysis (Partana) Tabdev is a method to assess the quality of partitions with respect to the original data, and not its distance matrix representation. See tabdev for details.

Value

Partana returns an S3 object of class "partana".

Tabdev returns an S3 object of class "tabdev", a list with two elements 'spcdev' and 'totdev'.

spcdev

data.frame with four columns giving the species ("numocc"), the number of occurrences of a species in the data set ("numocc"), estimated deviance "deviance" and probability (pval).

totdev

total tabele deviance.

Note

To prevent

Author(s)

Interfaced by Roland Kaiser for package vegsoup using David W. Roberts tabdev, partana, disdiam and silhouette algorithms in package optpart.

See Also

optsil, optindval

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
require(vegsoup)

data(barmstein)
x <- VegsoupPartition(barmstein, k = 2)

## we need to use :: because optpart does not properly register S3methods
require(optpart)
vegsoup::silhouette(x)
vegsoup::disdiam(x)
vegsoup::tabdev(x)
vegsoup::partana(x)

vegsoup documentation built on Feb. 24, 2021, 3 a.m.