glzcluster {octopucs}

Generalized Cluster

Description

Generates a cluster hypothesis based on n subsets of the available columns.

Usage

glzcluster(data, n = NULL, prop=NULL, distance =NULL, method=NULL, Name=NULL)

Arguments

data

A matrix object of r rows that will be grouped by c columns.

n

The number of subsets to be generated is based on which the generalized cluster will be generated.

prop

Either the proportion of the c columns used in every run to produce a new grouping array or the 1-prop number of cells in the body of the matrix whose values are replaced by the corresponding column average. This will be automatically decided based on the total number of columns in the matrix. When there are fewer than 19 columns in the matrix, the procedure will replace cell values with column averages even if the resample is TRUE.

distance

Distance method to be used (default="bray"). Options include those of the function "vegdist" ("vegan"): "manhattan", "euclidean", "canberra", "bray", "kulczynski", "gower", "morisita", "horn", "mountford", "jaccard", "raup", "binomial", "chao", "altGower", "cao" and "mahalanobis", as well as those referred in Koleff et al. (2003), available either by name or number: 1 or "w", 2 or "-1", 3 or "c", 4 or "wb", 5 or "r", 6 or "I", 7 or "e", 8 or "t", 9 or "me", 10 or "j", 11 or "sor", 12 or "m", 13 or "-2", 14 or "co", 15 or "cc", 16 or "g", 17 or "-3", 18 or "1", 19 or "", 20 or "hk", 21 or "rlb", 22 or "sim", 23 or "gl", 24 or "z". The axis is re-scaled to cover a zero-one range for open distance metrics (not restricted to zero-one ranges). If so, the axis will be labeled a "Similarity (rescaled)"; if no re-scaling is necessary, the label will appear as "Similarity".

method

Agglomerative procedure for hierarchical clustering, including: "average" (=UPGMA), "ward.D", "ward.D2", "single", "complete", "mcquitty" (=WPGMA), "median" (=WPGMC) or "centroid" (=UPGMC).

Name

Name of the object matrix analyzed

Value

Generates a list object that contains the n cluster assableged with random subsets of the data and in the n+1 possition the generalized cluster, an hclust output

Author(s)

Roger Guevara ([e-mail]) & Ian MacGregor-Fors ([e-mail])

References

Koleff, P., Gaston, K.J., Lennon, J. J. 2003. Measuring beta diversity for presence-absence data. Journal of Animal Ecology 71:367. https://doi.org/10.1046/j.1365-2656.2003.00710.x

Examples

data(dune)
M2 <- glzcluster(dune, n=100,
prop=0.8, distance="bray",
method="average")
length(M2)
ls(M2[[length(M2)]])