Information Technology Reference
In-Depth Information
0.5
l
V5
n
1.5
e
1
i
0.5
0.5
d
0.5
a
0
0
V7
0
0
m
c
P
V12
f
0.5
h
o
b
j
k
1
0.5
g
1.5
V9
2
2.5
Figure 5.28 The interpolative nonlinear biplot of the soils data as constructed in the
bottom left panel of Figure 5.27, illustrating the graphical interpolation of a point P with
values V5
=− 1 . 0, V7
=− 1 . 1, V9 = 2and V12
=− 0 . 5.
to the inter-sample distances than the other three variables. By way of illustration we
translate the data set by adding 0.2 to each value for each variable, thus resolving any
potential problems with a zero value, and then calculate the squared inter-sample Clark's
distances as given in Table 5.1.
The biplots for normal projection and circle projection are shown in Figures 5.29
and 5.30, respectively. The function call for constructing the biplot in Figure 5.29 is
Nonlinbipl(X = aircraft.data[,2:5]+0.2, ax.type = "predictive",
dist = "Clark", prediction.type = "normal", colours = "blue",
pch.samples = 15, num.points = 50)
Changing the argument prediction.type = "circle" leads to the biplot in
Figure 5.30. Although the biplot trajectories appear different, the different projection
methods result in equivalent predictions. In Figure 5.30 we show the circle predic-
tions for aircraft f and g by calling circle.projection.interactive(col =
"green", cent.line = TRUE) and selecting the required position. The function
Search WWH ::




Custom Search