Seriation heuristics for symmetric matrices

Description

Seriation heuristics for symmetric matrices used for cluproxplot. Seriation tries to move large values towards the main diagonal by reordering columns and rows at the same time.

Usage

seriation(x, method = NULL, args = NULL)

Arguments

Details

Currently the following methods are implemented:

Murtagh (1985), Algorithm B

a simple heuristic (default method).

Hierarchical Clustering

uses hierarchical clustering on the matrix; the method can be passed as the element method in args (defaults to "average"). The returned ordering is the ordering of the leaves in the resulting dendrogram.

Gruvaeus and Wainer (1972)

uses hierarchical clustering and reorders the leaves using a heuristic.

Optimal Leaf Ordering

uses hierarchical clustering and finds the optimal permutation of the leaves in the dendrogram.

Value

The order as a vector of integer. The attribute "method" contains the method as a character string.

Author(s)

Michael Hahsler (hahsler@ai.wu-wien.ac.at)

References

F. Murtagh (1985). Multidimensional Cluster Algorithms. Lectures in Computational Statistics, Physica Verlag, pp. 15.

Gruvaeus, G. and Wainer, H. (1972), "Two Additions to Hierarchical Cluster Analysis", British Journal of Mathematical and Statistical Psychology, 25, 200-206.

Hurley, Catherine B. Clustering Visualizations of Multidimensional Data. Journal of Computational and Graphical Statistics, 13(4), pages 788-806, 2004.

Z. Bar-Joseph, E. D. Demaine, D. K. Gifford, and T. Jaakkola. (2001) Fast Optimal Leaf Ordering for Hierarchical Clustering. Bioinformatics, Vol. 17 Suppl. 1, pp. 22-29.

See Also

Examples

data("Votes")

### create dummy coding (with removed party affiliation)
x <- as.dummy(Votes[-17])

### calculate distance matrix
d <- dists(x, method = "binary")

### use standard method
order <- seriation(d)
order

### use hclust w/complete linkage + optimal leaf ordering
order2 <- seriation(d, method = "Optimal")
order2

### use hclust w/single linkage + optimal leaf ordering
order3 <- seriation(d, method = "Optimal", args = list(method = "single"))
order3