Information Technology Reference
In-Depth Information
Hun(62)
50
150
Box.H
0
Cap(72)
Tru.H
50
50
50
100
Box.L
0
50
T64(30)
0
0
0
Tru.L
Ear.L
50
0
0
0
50
100
50
50
Hob(78)
20
Ear.H
Fow.H
50
0
Edn.H
Cra.L
0
Beg.L
50
100
0
Cra.H
50
Edn.L
0
50
100
Beg.H
50
150
Fow.L
Spo(44)
100
100
T95(8)
Kin(75)
100
100
50
Figure 6.12 Predicting the values (in terms of interactions with main effects added) for
all sites for the same variety ( Hob ). Accomplished by calling biadbipl with arguments
X = wheat.data, predictions.allsamples.onaxis = 8 .
The reader is encouraged to reproduce the three-dimensional biadditive biplots and to
use the mouse buttons to interactively explore the three-dimensional biplot display.
As a final exercise, the reader is asked to reproduce the two-dimensional biplots
of the interaction matrix of the wheat data with extra points added representing the
average yield of the high nitrogen and low nitrogen at each site together with the
predictivities of these extra points. First, read Section 6.5 carefully. Recall arguments
X.new.rows and X.new.columns of biadbipl . Recall the output of function biad.
predictivities .
If an estimate of the true error variance is available, biadditive biplots may also be
embellished with confidence ellipses. In this regard, the reader is referred to Gower and
Hand (1996) and the references contained therein.
6.8 Diagnostic biplots
The foregoing gives the most important use of biplots for a quantitative two-way table.
Another, less used, application is to diagnose what model to fit, chosen from within the
Search WWH ::




Custom Search