-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsimulation_analysis.R
295 lines (246 loc) · 8.19 KB
/
simulation_analysis.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
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
library(fda)
library(funFEM)
library(forecast)
library(funHDDC)
library(mclust)
library(cvTools)
library(clue)
source("./simulation_models.R")
source("./functions.R")
total.n.sim=100
TP_result1=TP_result2=matrix(nrow=5, ncol=total.n.sim)
TP_maj_result1<-TP_maj_result2<-vector()
TP_cv_result1=TP_cv_result2<-vector()
ccr.funhddc=ccr.funfem =ccr.dtw=vector()
ar.TP_result1=ar.TP_result2=matrix(nrow=5, ncol=total.n.sim)
ar.TP_maj_result1<-ar.TP_maj_result2<-vector()
ar.TP_cv_result1=ar.TP_cv_result2<-vector()
ar.funhddc=ar.funfem =ar.dtw=vector()
cv_opt=vector()
time_d<-matrix(nrow=total.n.sim, ncol=9)
model_type='1A'## 1A , 2A, 2B, block, 4A(sigma 0.1, 0.5)
###################################################
######### Simulation model figures ############
###################################################
n=100
TT=500
data_temp= data_generation(n=n, TT=TT, model=model_type)
print(length(which(data_temp[[1]]==0))/length(data_temp[[1]]) )
par(mfrow=c(1,2), mai=c(0.5,0.8,0.8,0.5))
for(i in c(1: length(data_temp)) ){
sam=sample(c(1:n), 1)
plot(data_temp[[i]][sam,], type='b',pch=20, cex=0.7, ylab='', xlab='',
main=paste('Model 1 (g=', i,')' , sep=''), ylim=c(0,7) )
abline(v=54,col=2,lty=2,lwd=2)
abline(v=128,col=2,lty=2,lwd=2)
}
###################################################
######### Cross Validation for Thickness ############
###################################################
for(sim in 1:total.n.sim){
print(paste(sim,'th interation'))
set.seed(sim)
n=100
TT=500
skip_sim <- FALSE
#Generate Dataset
data_temp= data_generation(n=n, TT=TT, model=model_type)
data=data_temp[[1]]
for(l in 2:length(data_temp)){
data=rbind(data ,data_temp[[l]] )
}
data=t(data) ##TT*(N=n*K) matrix
dim(data)
K=length(data_temp)
perm_num=nrow(perm(1:K))
true_cl=matrix(ncol=perm_num, nrow=n*K)
for(j in 1:perm_num){
true_cl[,j]=rep(perm(1:K)[j,] , each=n)
}
tryCatch({
start_time <- Sys.time()
## various t value ####
t1_index=c(10,20,30,50,100,150)
cv_error<-matrix(nrow=2, ncol=length(t1_index))
for(k in 1:length(t1_index)){
t1=t1_index[k]
r1=0.1
## Cross-validation for thickness code ####
K_cv=5
folds <- cvFolds(ncol(data), K=K_cv)
temp_error_ccr<-temp_error_adj<-vector()
for(i in 1:K_cv){
train <- data[,folds$subsets[folds$which != i] ] #Set the training set
validation <- data[,folds$subsets[folds$which == i] ] #Set the validation set
Up_sim = prepare_up(train,FUN=sqBound_es,t1,r1)
a1= cluster_zits(K=K,Up=log(Up_sim), FUN=l1_mean ,t1,r1,is.median=T)
Up_test = prepare_up(validation,FUN=sqBound_es,t1,r1)
pmeasure=matrix(nrow=ncol(Up_test) , ncol=K)
for(clstr in 1:K){
for(j in 1:ncol(Up_test)){
pmeasure[j,clstr] = l1_mean(a1$pivot[,clstr],log(Up_test[,j]))
}
}
yhat=apply(pmeasure,1, which.min)
#assign each validation set to the pivots derived using train set
true_cl2=true_cl[ folds$subsets[folds$which == i] ,]
ccr=vector()
for(j in 1:perm_num){
ccr[j]=sum(diag(table(true_cl2[,j],yhat)))/(length(yhat))
}
temp_error_ccr[i]=max(ccr)
temp_error_adj[i]=adjustedRandIndex(true_cl2[,1],yhat)
}
cv_error[1,k]=mean(temp_error_adj)
cv_error[2,k]=mean(temp_error_ccr)
}
#par(mfrow=c(1,2))
#plot(t1_index, cv_error[1,], ylab='Adj. rand index', xlab='thickness', type='b')
#plot(t1_index, cv_error[2,], type='b', ylab='CCR', xlab='thickness' )
#Optimal Thickness based on 5 fold CV.
opt_k=which.max( cv_error[2,])
t1=t1_index[opt_k]
print(t1)
cv_opt[sim]=t1
r1=0.1
Up_sim = prepare_up(data,FUN=sqBound_es,t1,r1)
a1= cluster_zits(K=K,Up=log(Up_sim), FUN=l1_mean ,t1,r1,is.median=1)
yhat=vector()
for(l in 1:K ){
yhat[a1$clusters[[l]]]=l
}
ccr=vector()
for(j in 1:perm_num){
ccr[j]=sum(diag(table(true_cl[,j],yhat)))/(ncol(data))
}
TP_cv_result1[sim]=max(ccr)
ar.TP_cv_result1[sim]= adjustedRandIndex(true_cl[,1],yhat)
end_time <- Sys.time()
time_d[sim,6 ] = round(difftime(end_time, start_time, units = "secs"), 3)
}, error = function(e) {
print("TP error")
print(e)
skip_sim <<- TRUE
})
if (skip_sim == TRUE) {
next
}
}
###################################################
######### Simulation running 100 times ############
###################################################
#### Use parallel computing and compute time
library(parallel)
nCores = detectCores()
s <- system.time({
run_sim <- mclapply(1:total.n.sim, function(sim){
print(paste(sim,'th interation'))
set.seed(sim)
n=100
TT=500
#Generate Dataset
data_temp= data_generation(n=n, TT=TT, model=model_type)
data=data_temp[[1]]
for(l in 2:length(data_temp)){
data=rbind(data ,data_temp[[l]] )
}
data=t(data) ##TT*(N=n*K) matrix
dim(data)
K=length(data_temp)
perm_num=nrow(perm(1:K))
true_cl=matrix(ncol=perm_num, nrow=n*K)
for(j in 1:perm_num){
true_cl[,j]=rep(perm(1:K)[j,] , each=n)
}
tryCatch({
start_time <- Sys.time()
t1_index=c(10,20,30,50,100,150)
#par(mfrow=c(1,2))
#plot(t1_index, cv_error[1,], ylab='Adj. rand index', xlab='thickness', type='b')
#plot(t1_index, cv_error[2,], type='b', ylab='CCR', xlab='thickness' )
opt_k=4
t1=t1_index[opt_k]
print(t1)
#cv_opt[sim]=t1
r1=0.1
Up_sim = prepare_up(data,FUN=sqBound_es,t1,r1)
a1= cluster_zits(K=K,Up=log(Up_sim), FUN=l1_mean ,t1,r1,is.median=1)
yhat=vector()
for(l in 1:K ){
yhat[a1$clusters[[l]]]=l
}
ccr=vector()
for(j in 1:perm_num){
ccr[j]=sum(diag(table(true_cl[,j],yhat)))/(ncol(data))
}
end_time <- Sys.time()
return(list(obj = a1$obj ,ccr = max(ccr),aRand = adjustedRandIndex(true_cl[,1],yhat), time =round(difftime(end_time, start_time, units = "secs"), 3) ))
}, error = function(e) {
print("TP error")
print(e)
skip_sim <<- TRUE
})
if (skip_sim == TRUE) {
next
}
}, mc.cores = nCores-1)
})
run_sim1 <- unlist(run_sim)
run_sim1 <- matrix(run_sim1,ncol=4,byrow = T)
colnames(run_sim1) <- c("obj","ccr","arand","time")
run_sim1 = as_tibble(run_sim1)
###################################################
######## Comparison methods running 100 times #######
###################################################
library(foreach)
library(doParallel)
nCores = detectCores()
cl <-makeCluster(nCores-1)
registerDoParallel(cl)
funfem_result<-foreach(sim = 1:total.n.sim, .combine = rbind) %dopar% {
set.seed(sim)
n=100
TT=1500
#Generate Dataset
data_temp= data_generation(n=n, TT=TT, model=model_type)
data=data_temp[[1]]
for(l in 2:length(data_temp)){
data=rbind(data ,data_temp[[l]] )
}
data=t(data) ##TT*(N=n*K) matrix
K=length(data_temp)
perm_num=nrow(perm(1:K))
true_cl=matrix(ncol=perm_num, nrow=n*K)
for(j in 1:perm_num){
true_cl[,j]=rep(perm(1:K)[j,] , each=n)
}
start_time <- Sys.time()
basis <- create.fourier.basis(c(0, nrow(data)), nbasis=11)
fdobj <- smooth.basis(1:nrow(data), data,basis)$fd
cl.sinu = funFEM(fdobj, K=K, init="kmeans", model="all")
elements = list()
for(c in 1:K){
# get the point of cluster c
elements[[c]] <- which(cl.sinu$cls == c)
}
yhat=vector()
for(l in 1:K ){
yhat[elements[[l]]]=l
}
#######################################
###### funHDDC
# res.uni <- funHDDC(fdobj,K=K,model="AkBkQkDk",init="kmeans",threshold=0.2)
# yhat= res.uni$class
###### DTW
#pc.l2 <-tsclust(t(data), k = K, distance = "sdtw", centroid = "pam",seed = 3247, trace = FALSE, control = partitional_control(nrep = 5L))
#yhat=cl_class_ids(cl_medoid(pc.l2))
#######################################
#Compute CCR
ccr=vector()
for(j in 1:perm_num){
ccr[j]=sum(diag(table(true_cl[,j],yhat)))/(ncol(data))
}
end_time <- Sys.time()
(data.frame(ccr.funfem = max(ccr), ar.funfem= adjustedRandIndex(true_cl[,1],yhat), time = round(difftime(end_time, start_time, units = "secs"), 3)))
}
stopCluster(cl)