Environmental Engineering Reference
In-Depth Information
Table 1.11
Mahalanobis distances for the dataset in
Table 1.9
k
X
1
−
m
1
X
2
−
m
2
X
3
−
m
3
Q
d
1
1.11
0.28
1.62
−
0.92
2
0.58
−
0.98
0.33
1.00
3
1.65
0.87
4.87
−
0.026
4
0.69
−
0.57
−
0.11
−
0.38
5
1.59
−
1.87
1.09
3.42
6
−
1.37
−
0.12
−
1.15
3.14
7
1.20
1.33
−
1.03
−
0.59
8
0.18
0.50
0.95
2.23
9
1.12
5.02
−
1.52
−
1.99
10
−
0.60
1.21
0.59
3.67
−
[
−
n
q
(
)]
on the
2
χ
d
−
−[ ( )] rela-
tionship lies close to the 1:1 line, we can conclude that there is no strong evidence to reject
the multivariate normal model.
Table 1.12
illustrates how the line test can be applied to simulated multivariate standard
ing order. The second column contains the rank. The third column computes the ECDF
horizontal axis will produce a 1:1 line. As a result, if the actual
q
d
versus F1 F
χ
d
n
q
2
−
−[ ( )],
where F
n
(
q
d
) is the data in
the third column. Finally, the probability plot is obtained by drawing the first column on
F1 F
χ
d
n
q
2
χ
d
−
−[ ( )],
relationship. It is indeed close to the 1:1 line, indicating the multivariate
normal model is reasonable.
n
q
2
1.4.4 Simulation of multivariate standard normal
random vector X
Given the correlation matrix
C
, random samples of multivariate standard normal (X
1
, X
2
,
…, X
d
) can be readily simulated by the following steps:
1. Simulate independent standard normal random vector
Z
= [Z
1
, Z
2
, …, Z
d
]
T
.
Table 1.12
Sorted Q
d
data and the
EDF
Sorted Q
d
data
Rank k
F[1F(q )]
2
−
1
−
nd
χ
d
0.69
1
0.067
0.44
1.00
2
0.163
0.85
1.33
3
0.260
1.25
1.62
4
0.356
1.67
2.23
5
0.452
2.12
3.14
6
0.548
2.63
3.42
7
0.644
3.24
3.67
8
0.740
4.02
4.87
9
0.837
5.12
5.02
10
0.933
7.15
Search WWH ::
Custom Search