-
Notifications
You must be signed in to change notification settings - Fork 0
/
vivj_matrix.R
198 lines (161 loc) · 6.93 KB
/
vivj_matrix.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
# Select any two species pair [i,j] for a copula of location loc
# This function gives you a matrix with vi and vj as two columns
# Input :
# d_allsp : dataset in data[[loc]][[sp]] format
# loc : location index
# i,j : sp-pair indices
# level : significance level for BiCopIndepTest p-value
# ploton : (optional) logical, if T gives copula plot without transforming j-th variable to it's -ve value
# onbounds : a logical tag (default=FALSE) to get info about the points exactly lying on bounds, if set to TRUE
# then the arguments lb and ub must be numeric
# lb : numeric value between [0,1] for lower bound (default =NA)
# ub : numeric value between [0,1] for upper bound (default =NA), ub should be greater than lb
# include_indep: logical (whether to keep track for indep. test or not)
# Output :
# A list of 4 elements:
# mat : a matrix : copula of (vi,vj) with transforming j-th variable to it's -ve value for -ve corr.
# corval : Spearman's correlation
# pval : pvalue of Spearman's cor.test
# IndepTestRes : BiCopIndepTest p-value
# and an optional plot of the copula
library(VineCopula)
vivj_matrix<-function(d_allsp,loc,i,j,level=0.05,ploton,onbounds=F,lb=NA,ub=NA,include_indep){
ds1<-d_allsp[[loc]][[i]]
ds2<-d_allsp[[loc]][[j]]
#----------------------------
colnames(ds1)<-c("Year","Dat") # ensuring column names
colnames(ds2)<-c("Year","Dat")
a1<-ds1$Year[1]
a2<-ds2$Year[1]
a3<-ds1$Year[dim(ds1)[1]]
a4<-ds2$Year[dim(ds2)[1]]
year_s<-max(a1,a2)
year_e<-min(a3,a4)
ind_s1<-which(ds1$Year==year_s)
ind_s2<-which(ds2$Year==year_s)
ind_e1<-which(ds1$Year==year_e)
ind_e2<-which(ds2$Year==year_e)
ds1<-ds1[ind_s1:ind_e1,]
ds2<-ds2[ind_s2:ind_e2,]
# Omitting the years and data containing NA in either d1 or d2
#from both d1 and d2
if(anyNA(ds1$Dat)==T | anyNA(ds2$Dat)==T){
ind_na1<-which(is.na(ds1$Dat))
ind_na2<-c(ind_na1,which(is.na(ds2$Dat)))
ind_na<-unique(ind_na2)
d1Dat<-ds1$Dat[-ind_na]
d2Dat<-ds2$Dat[-ind_na]
Years<-ds1$Year[-ind_na]
d1<-data.frame(Year=Years,Dat=d1Dat)
d2<-data.frame(Year=Years,Dat=d2Dat)
}else{
d1<-ds1
d2<-ds2
}
colnames(d1)[2]<-"Dat" # ensuring column names
colnames(d2)[2]<-"Dat"
#get ranks modified now
vi<-VineCopula::pobs(d1$Dat)
vj<-VineCopula::pobs(d2$Dat)
IndepTestRes<-VineCopula::BiCopIndTest(vi,vj)$p.value
ct<-cor.test(vi,vj,alternative = "two.sided",method="spearman",exact=F)
corval<-unname(ct$estimate)
pval<-ct$p.value
if(ploton==T){
if(include_indep==T){
if(IndepTestRes<level && corval>0){ # for significant positive correlation
plot(vi,vj,type='p',col=rgb(0,0,0,0.3),pch=19,xlim=c(0,1),ylim=c(0,1),
xlab=names(d_allsp[[loc]])[i],ylab=names(d_allsp[[loc]])[j],cex.lab=1.5)
if(j>i){
if(onbounds==T & identical(vi,vj)==F){
ind_lb<-which(vi+vj==(2*lb))
ind_ub<-which(vi+vj==(2*ub))
onlb<-length(ind_lb)
onub<-length(ind_ub)
if(onlb!=0 | onub!=0){
mtext(paste0("onbs = (",onlb," , ",onub,")"),
side = 4, line=0.15, adj=0.5, col="red")
}
}
}
}else if(IndepTestRes<level && corval<0){ # for significant negative correlation
plot(vi,vj,type='p',col=rgb(0,1,0,0.3),pch=19,xlim=c(0,1),ylim=c(0,1),
xlab=names(d_allsp[[loc]])[i],ylab=names(d_allsp[[loc]])[j],cex.lab=1.5)
if(j>i){
if(onbounds==T & identical(vi,vj)==F){
vneg<-VineCopula::pobs(-(d2$Dat)) # see when we count points on bounds we took reverse of second variable
ind_lb<-which(vi+vneg==(2*lb))
ind_ub<-which(vi+vneg==(2*ub))
#vneg<-VineCopula::pobs(-(d1$Dat)) # NOTE : onbs will not be same if we consider first variable to be reversed
#ind_lb<-which(vj+vneg==(2*lb))
#ind_ub<-which(vj+vneg==(2*ub))
onlb<-length(ind_lb)
onub<-length(ind_ub)
if(onlb!=0 | onub!=0){
mtext(paste0("onbs = (",onlb," , ",onub,")"),
side = 4, line=0.15, adj=0.5, col="red")
}
}
}
}else{ # independent case
plot(-1,0,xlim=c(0,1),ylim=c(0,1),xlab=names(d_allsp[[loc]])[i],ylab=names(d_allsp[[loc]])[j],cex.lab=1.5)
text(0.5,0.5,"Indep.",adj=c(0.5,.5),cex=2)
}
mtext(paste0("(sp_x, sp_y) = (",i," , ",j,")"),
side = 3, line=0.15, adj=0.5, col="black")
if(IndepTestRes<level && corval<0){
vj<-VineCopula::pobs(-(d2$Dat))
}
}else{ # when we consider all cells however week their correlations were.
if(corval>0){ # for all +ve correlations
plot(vi,vj,type='p',col=rgb(0,0,0,0.3),pch=19,xlim=c(0,1),ylim=c(0,1),
xlab=names(d_allsp[[loc]])[i],ylab=names(d_allsp[[loc]])[j],cex.lab=1.5)
if(j>i){
if(onbounds==T & identical(vi,vj)==F){
ind_lb<-which(vi+vj==(2*lb))
ind_ub<-which(vi+vj==(2*ub))
onlb<-length(ind_lb)
onub<-length(ind_ub)
if(onlb!=0 | onub!=0){
mtext(paste0("onbs = (",onlb," , ",onub,")"),
side = 4, line=0.15, adj=0.5, col="red")
}
}
}
}else{ # for all -ve correlations
plot(vi,vj,type='p',col=rgb(0,1,0,0.3),pch=19,xlim=c(0,1),ylim=c(0,1),
xlab=names(d_allsp[[loc]])[i],ylab=names(d_allsp[[loc]])[j],cex.lab=1.5)
if(j>i){
if(onbounds==T & identical(vi,vj)==F){
vneg<-VineCopula::pobs(-(d2$Dat)) # see when we count points on bounds we took reverse of second variable
ind_lb<-which(vi+vneg==(2*lb))
ind_ub<-which(vi+vneg==(2*ub))
#vneg<-VineCopula::pobs(-(d1$Dat)) # NOTE : onbs will not be same if we consider first variable to be reversed
#ind_lb<-which(vj+vneg==(2*lb))
#ind_ub<-which(vj+vneg==(2*ub))
onlb<-length(ind_lb)
onub<-length(ind_ub)
if(onlb!=0 | onub!=0){
mtext(paste0("onbs = (",onlb," , ",onub,")"),
side = 4, line=0.15, adj=0.5, col="red")
}
}
}
}
if(corval<0){ #reverse the variable
vj<-VineCopula::pobs(-(d2$Dat))
}
}
}
Years<-d1$Year
#-------------------------
#n_datapt<-length(vi)
#--------------------
#plot(vi,vj,type="p")
#-------------------------
mat<-as.matrix(cbind(vi,vj))
return(list(mat=mat, # return reversed mat so that if you plot this mat you get +ve correlation
corval=corval, # but return the actual -ve corr. value
pval=pval,
IndepTestRes=IndepTestRes))
}