Plot a Web of Connected Points

Share:

Description

This plot can be used to graphically display a correlation matrix by using the linewidth between the nodes in proportion to the correlation of two variables. It will place the elements homogenously around a circle and draw connecting lines between the points.

Usage

1
2
3
PlotWeb(m, col = c(hred, hblue), lty = par("lty"), lwd = NULL, args.legend=NULL,
        pch = 21, pt.cex = 2, pt.col = "black", pt.bg = "darkgrey",
        cex.lab = 1, las = 1, adj = NULL, dist = 0.5, ...)

Arguments

m

a symmetric matrix of numeric values

col

the color for the connecting lines

lty

the line type for the connecting lines

lwd

the line widths for the connecting lines. If left to NULL it will be linearly scaled between the minimum and maximum value of m.

args.legend

list of additional arguments to be passed to the legend function. Use args.legend = NA if no legend should be added.

pch

the plotting symbols appearing in the plot, as a non-negative numeric vector (see points, but unlike there negative values are omitted) or a vector of 1-character strings, or one multi-character string.

pt.cex

expansion factor(s) for the points.

pt.col

the foreground color for the points, corresponding to its argument col.

pt.bg

the background color for the points, corresponding to its argument bg.

las

alignment of the labels, 1 means horizontal, 2 radial and 3 vertical.

adj

adjustments for the labels. (Left: 0, Right: 1, Mid: 0.5)

dist

gives the distance of the labels from the outer circle. Default is 2.

cex.lab

the character extension for the labels.

...

dots are passed to plot.

Value

A list of x and y coordinates, giving the coordinates of all the points drawn, useful for adding to the graph.

Author(s)

Andri Signorell <andri@signorell.net>

See Also

PlotCorr

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
m <- cor(d.pizza[, which(sapply(d.pizza, IsNumeric, na.rm=TRUE))[-c(1:2)]],
         use="pairwise.complete.obs")
PlotWeb(m=m, col=c(hred, hblue), main="Pizza Correlation")


# let's describe only the significant corrs and start with a dataset
d.m <- d.pizza[, which(sapply(d.pizza, IsNumeric, na.rm=TRUE))[-c(1:2)]]

# get the correlation matrix
m <- cor(d.m, use="pairwise.complete.obs")

# let's get rid of all non significant correlations
ctest <- PairApply(d.m,  function(x, y) cor.test(x, y)$p.value, symmetric=TRUE)

# ok, got all the p-values, now replace > 0.05 with NAs
m[ctest > 0.05] <- NA
# How does that look like now?
noquote(Format(m, na.form = ". ", leading="drop", digits=3, align = "right"))

PlotWeb(m, las=2, cex=1.2)

# define line widths
PlotWeb(m, lwd=abs(m[lower.tri(m)] * 10))

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.