shipunov (version 1.17.1)

PlotBest.mdist: Dotchart which reflects the "best" of non-base distances

Description

Plots dotchart which shows correspondences between data and various non-base distances

Usage

PlotBest.mdist(data, distances=c("manhattan", "euclidean", "canberra",
"clark",  "bray", "kulczynski", "jaccard", "gower", "altGower",
"morisita", "horn", "binomial",  "chao", "cao", "mahalanobis",
"cor.pearson", "cor.spearman", "cor.kendall",  "gower_dist",
"simple_match_dist", "daisy.gower", "smirnov"),
binary.only=FALSE)

Arguments

data

Data frame with values

distances

Distances to use

binary.only

Use binary only distances?

Author

Alexey Shipunov

Details

Shows the "best" distance method using many non-base distances from several packages (namely, "cluster", "smirnov" and "vegan" -- but does not include "mountford" and "raup" as they are very special). Please note that this is a mere visualization, and numbers are used only to understand the relative correspondence between raw data and distances.

Uses maximal correlations between multidimensional scaling of distance object (converted internally to Euclidean) and PCA of data. Both MDS and PCA use two dimensions.

See Also

PlotBest.dist

Examples

PlotBest.mdist(iris[, -5])

m1 <- t((moldino > 0) * 1)
PlotBest.mdist(m1, binary.only=TRUE)