Page 159 - Computational Statistics Handbook with MATLAB

146 Computational Statistics Handbook with MATLAB

vates the same capability. One reason for looking at scatterplots of the data is
to look for interesting structures. The ability to view these structures for 3-D
data is dependent on the viewpoint or projection to the screen. When looking
at 3-D scatterplots, the analyst should rotate them to search the data for pat-
terns or structure.

Example 5.16
Three variables were measured on ten insects from each of three species
[Hand, et al.,1994]. The variables correspond to the width of the first joint of
the first tarsus, the width of the first joint of the second tarsus and the maxi-
mal width of the aedeagus. All widths are measured in microns. These data
were originally used in cluster analysis [Lindsey, Herzberg, and Watts, 1987].
What we would like to see from the scatterplot is whether the data for each
species can be separated from the others. In other words, is there clear sepa-
ration or clustering between the species using these variables? The 3-D scat-
terplot for these data is shown in Figure 5.25. This view of the scatterplot
indicates that using these variables for pattern recognition or clustering (see
Chapter 9) is reasonable.

% Load the insect data
load insect
% Create a 3-D scatter plot using a
% different color and marker
% for each class of insect.
% Plot the first class and hold the plot.
plot3(insect(1:10,1),insect(1:10,2),...
insect(1:10,3),'ro')
hold on
% Plot the second class.
plot3(insect(11:20,1),insect(11:20,2),...
insect(11:20,3),'gx')
% Plot the third class.
plot3(insect(21:30,1),insect(21:30,2),...
insect(21:30,3),'b*')
% Be sure to turn the hold off!
hold off

© 2002 by Chapman & Hall/CRC