-
Notifications
You must be signed in to change notification settings - Fork 0
/
Outlier increasing.R
211 lines (190 loc) · 6.19 KB
/
Outlier increasing.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
##Outlier increasing================
library(MASS)
library(rrcov)
library(MASS)
library(robustbase)
library(robust)
library(pcaPP)
library(mvtnorm)
library(clusterGeneration)
#=====================Data Generation ========================
set.seed(0001)
mu=rep(0,6)
v=genPositiveDefMat("unifcorrmat",dim=6)$Sigma
Data=mvrnorm(n=100,mu,v);colnames(Data)=c("x1","x2","x3","y1","y2","y3")
#========================Correlation Matrix=========================
set.seed(002)
conta=matrix(rep(round(5*as.numeric(apply(Data,2,max))+rnorm(6,0,1)),50),50)
CData=rbind(Data[-c(26:75),],conta)
#dim(Data)
#===================================================================
##==============Different Covariance matrix========================
#===================================================================
# We observe that there also same result though the data are contaminated
##========================================================================
v.bve<-cov2cor(MeanVar(CData,Beta=0.1)$V)
v.mve<-cov2cor(CovMve(CData)@cov)
v.mcd<-cov2cor(CovMcd(CData)@cov) #?CovMcd
v.ogk<-cov2cor(covOGK(CData, sigmamu = scaleTau2)$wcov)
#=======================================================================
##=============Robustification by Beta divergence ================
##======================================================================
r11<-v.bve[1:3,1:3]
r22<-v.bve[4:6,4:6]
r12<-v.bve[1:3,4:6]
r21<-t(r12)
#===================== squre root matrix for r11===================
ev1<-eigen(r11)
sev1<-sqrt(ev1$values)
dm1<-diag(1/sev1)
sr11<-ev1$vectors%*%dm1%*%t(ev1$vectors)
#===================== squre root matrix for r22====================
ev2<-eigen(r22)
sev2<-sqrt(ev2$values)
dm2<-diag(1/sev2)
sr22<-ev2$vectors%*%dm2%*%t(ev2$vectors)
#====================== Define Cross correlation matrix m ==========
m<-sr11%*%r12%*%sr22
#====================== compute SVD OF m ===========================
sd<-svd(m)
#====================== canonical correlation========================
can.corr1<-sd$d;can.corr1
l=can.corr^2
################P.Value Calculation of canonical variate paires##########
p=ncol(r11);q=ncol(r22);n=dim(Data)[1]
L=PV=df=matrix(rep(0,min(p,q)),nrow=min(p,q))
for(m in 1:min(p,q))
{
df[m]<-(p-m+1)*(q-m+1)
df
X=0
for(k in m:min(p,q))
{
X=X+log(1-l[k])
}
L[m]=-((n-1)-.5*(p+q+1))*X
PV=pchisq(L,df=df,lower.tail=F)
}
#L;PV;df
Table1=round(data.frame(CanCorr=can.corr1,Chisquare=L,df=df,p.value=PV),3);Table1
##=============Robustification by Mve estimator ================
##===========================================================================
r11<-v.mve[1:3,1:3]
r22<-v.mve[4:6,4:6]
r12<-v.mve[1:3,4:6]
r21<-t(r12)
#===================== squre root matrix for r11===================
ev1<-eigen(r11)
sev1<-sqrt(ev1$values)
dm1<-diag(1/sev1)
sr11<-ev1$vectors%*%dm1%*%t(ev1$vectors)
#===================== squre root matrix for r22====================
ev2<-eigen(r22)
sev2<-sqrt(ev2$values)
dm2<-diag(1/sev2)
sr22<-ev2$vectors%*%dm2%*%t(ev2$vectors)
#====================== Define Cross correlation matrix m ==========
m<-sr11%*%r12%*%sr22
#====================== compute SVD OF m ===========================
sd<-svd(m)
#====================== canonical correlation========================
can.corr2<-sd$d;can.corr2
l=can.corr^2
################P.Value Calculation of canonical variate paires##########
p=ncol(r11);q=ncol(r22);n=dim(Data)[1]
L=PV=df=matrix(rep(0,min(p,q)),nrow=min(p,q))
for(m in 1:min(p,q))
{
df[m]<-(p-m+1)*(q-m+1)
df
X=0
for(k in m:min(p,q))
{
X=X+log(1-l[k])
}
L[m]=-((n-1)-.5*(p+q+1))*X
PV=pchisq(L,df=df,lower.tail=F)
}
#L;PV;df
Table1=round(data.frame(CanCorr=can.corr2,Chisquare=L,df=df,p.value=PV),3);Table1
##=============Robustification by MCD estimator ================
##===========================================================================
r11<-v.mcd[1:3,1:3]
r22<-v.mcd[4:6,4:6]
r12<-v.mcd[1:3,4:6]
r21<-t(r12)
#===================== squre root matrix for r11===================
ev1<-eigen(r11)
sev1<-sqrt(ev1$values)
dm1<-diag(1/sev1)
sr11<-ev1$vectors%*%dm1%*%t(ev1$vectors)
#===================== squre root matrix for r22====================
ev2<-eigen(r22)
sev2<-sqrt(ev2$values)
dm2<-diag(1/sev2)
sr22<-ev2$vectors%*%dm2%*%t(ev2$vectors)
#====================== Define Cross correlation matrix m ==========
m<-sr11%*%r12%*%sr22
#====================== compute SVD OF m ===========================
sd<-svd(m)
#====================== canonical correlation========================
can.corr3<-sd$d;can.corr3
l=can.corr^2
################P.Value Calculation of canonical variate paires##########
p=ncol(r11);q=ncol(r22);n=dim(Data)[1]
L=PV=df=matrix(rep(0,min(p,q)),nrow=min(p,q))
for(m in 1:min(p,q))
{
df[m]<-(p-m+1)*(q-m+1)
df
X=0
for(k in m:min(p,q))
{
X=X+log(1-l[k])
}
L[m]=-((n-1)-.5*(p+q+1))*X
PV=pchisq(L,df=df,lower.tail=F)
}
#L;PV;df
Table1=round(data.frame(CanCorr=can.corr3,Chisquare=L,df=df,p.value=PV),3);Table1
#===========Robustification by OGK estimator=================
##========================================================================
r11<-v.ogk[1:3,1:3]
r22<-v.ogk[4:6,4:6]
r12<-v.ogk[1:3,4:6]
r21<-t(r12)
#===================== squre root matrix for r11===================
ev1<-eigen(r11)
sev1<-sqrt(ev1$values)
dm1<-diag(1/sev1)
sr11<-ev1$vectors%*%dm1%*%t(ev1$vectors)
#===================== squre root matrix for r22====================
ev2<-eigen(r22)
sev2<-sqrt(ev2$values)
dm2<-diag(1/sev2)
sr22<-ev2$vectors%*%dm2%*%t(ev2$vectors)
#====================== Define Cross correlation matrix m ==========
m<-sr11%*%r12%*%sr22
#====================== compute SVD OF m ===========================
sd<-svd(m)
#====================== canonical correlation========================
can.corr4<-sd$d ;can.corr4
l=can.corr^2
################P.Value Calculation of canonical variate paires##########
p=ncol(r11);q=ncol(r22);n=dim(Data)[1]
L=PV=df=matrix(rep(0,min(p,q)),nrow=min(p,q))
for(m in 1:min(p,q))
{
df[m]<-(p-m+1)*(q-m+1)
df
X=0
for(k in m:min(p,q))
{
X=X+log(1-l[k])
}
L[m]=-((n-1)-.5*(p+q+1))*X
PV=pchisq(L,df=df,lower.tail=F)
}
#L;PV;df
Table1=round(data.frame(CanCorr=can.corr4,Chisquare=L,df=df,p.value=PV),3);Table1
###return(list(Bcca=cancorr1,MVE=cancorr2,MCD=cancorr3,OGK=cancorr4))