scatterhist - Scatter plot with marginal histograms

Syntax

scatterhist(x,y) scatterhist(x,y,nbins) h = scatterhist(...)

Description

scatterhist(x,y) creates a 2-D scatterplot of the data in the vectors x and y, and puts a univariate histogram on the horizontal and vertical axes of the plot. x and y must be the same length.

The function is useful for viewing properties of random samples produced by functions such as copularnd, mvnrnd, lhsdesign.

scatterhist(x,y,nbins) also accepts a two-element vector nbins specifying the number of bins for the x and y histograms. The default is to compute the number of bins using a Scott rule based on the sample standard deviation. Any NaN values in either x or y are treated as missing, and are removed from both x and y. Therefore the plots reflect points for which neither x nor y has a missing value.

h = scatterhist(...) returns a vector of three axes handles for the scatterplot, the histogram along the horizontal axis, and the histogram along the vertical axis, respectively.

Examples

Example 1

Independent normal and lognormal random samples:

x = randn(1000,1);
y = exp(.5*randn(1000,1));
scatterhist(x,y)

Example 2

Marginal uniform samples that are not independent:

u = copularnd('Gaussian',.8,1000);
scatterhist(u(:,1),u(:,2))

Example 3

Mixed discrete and continuous data:

cars = load('carsmall');
scatterhist(cars.Weight,cars.Cylinders,[10 3])