Information Technology Reference
In-Depth Information
Ta b l e 4 . 2 0
Within-class axis predictivities (weighted and unweighted CVA) obtained
with CVA biplots of the copper froth data.
Weighted CVA
X1
X2
X3
X4
X5
X6
X7
X8
Dim_1
0.1005
0.0240
0.1891
0.0317
0.0319
0.0332
0.0627
0.0055
Dim_2
0.1005
0.3654
0.4602
0.8144
0.3350
0.3669
0.4312
0.0161
Dim_3
0.3517
0.6419
0.4675
0.8328
0.3492
0.3838
0.4449
0.1054
Dim_4
0.3525
0.7071
0.4706
0.8965
0.3842
0.4228
0.5040
0.9667
Dim_5
0.3546
0.7462
0.4707
0.9025
0.3904
0.4325
0.5052
0.9967
Dim_6
0.5458
0.7966
0.4936
0.9242
0.9940
0.9973
0.5963
0.9973
Dim_7
0.9705
0.8640
0.8053
0.9650
0.9984
0.9995
0.8217
0.9987
Dim_8
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
Unweighted
I
CVA
X1
X2
X3
X4
X5
X6
X7
X8
Dim_1
0.0957
0.0297
0.1709
0.0206
0.0235
0.0242
0.0508
0.0071
Dim_2
0.1000
0.2644
0.4546
0.7203
0.3177
0.3483
0.3737
0.0644
Dim_3
0.3508
0.5929
0.4560
0.7661
0.3228
0.3547
0.4008
0.1208
Dim_4
0.3525
0.7071
0.4706
0.8965
0.3842
0.4228
0.5040
0.9667
Dim_5
0.4937
0.7075
0.5159
0.9369
0.7796
0.7757
0.5395
0.9764
Dim_6
0.5599
0.7293
0.5911
0.9428
0.9722
0.9667
0.5442
0.9992
Dim_7
0.5778
0.9617
0.8293
0.9776
1.0000
0.9998
0.8713
1.0000
Dim_8
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
Unweighted and centred CVA
X1
X2
X3
X4
X5
X6
X7
X8
Dim_1
0.0957
0.0298
0.1706
0.0204
0.0233
0.0240
0.0506
0.0071
Dim_2
0.0991
0.2654
0.4507
0.7114
0.3100
0.3398
0.3679
0.0731
Dim_3
0.3524
0.6080
0.4522
0.7612
0.3142
0.3450
0.3987
0.1605
Dim_4
0.3525
0.7071
0.4706
0.8965
0.3842
0.4228
0.5040
0.9667
Dim_5
0.5780
0.7090
0.5337
0.9408
0.8390
0.8312
0.5350
0.9737
Dim_6
0.5794
0.8646
0.5508
0.9472
0.9931
0.9955
0.6200
0.9883
Dim_7
0.5890
0.9819
0.7981
0.9809
0.9957
0.9970
0.8534
0.9953
Dim_8
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
1.0000
4.11 Overlap in two-dimensional biplots
4.11.1 Describing the structure of overlap
In some instances it is not possible to completely separate the classes, as we saw
in the five-group example above. All is not lost, however. With
-bags we can still
deduce valuable conclusions even without separating the classes completely. In the
data given as
archaeology.data
in the R library we have measurements on Stone
Age artefacts for stone tools known as points. The complete data set of points and
other tools known as blades is discussed in detail in Wurz
et al.
(2003). The data
set contains three classes labelled
MSAI
,
MSAIIL
and
MSAIIU
indicating different
time
α
periods
in
the
Middle
Stone
Age.
The
oldest
artefacts
come
from
stage
I,
