-
Notifications
You must be signed in to change notification settings - Fork 11
/
arima_garch.r
192 lines (147 loc) · 5.8 KB
/
arima_garch.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
#"quantmod---------------------------------------------------------------------"
#install.packages("quantmod", repos='https://ftp.acc.umu.se/mirror/CRAN/')
#"lattice---------------------------------------------------------------------"
#install.packages("lattice", repos='https://ftp.acc.umu.se/mirror/CRAN/')
#"timeSeries---------------------------------------------------------------------"
#install.packages("timeSeries", repos='https://ftp.acc.umu.se/mirror/CRAN/')
#"rugarch---------------------------------------------------------------------"
#install.packages("rugarch", repos='https://ftp.acc.umu.se/mirror/CRAN/')
"Packages installed."
# defining a function
is.installed <- function(mypkg) is.element(mypkg, installed.packages()[,1])
checkNecessaryPackages <- function(){
"timeSeries:"
library("timeSeries")
is.installed('timeSeries')
"quantmod:"
library("quantmod")
is.installed('quantmod')
"lattice:"
library("lattice")
is.installed('lattice')
"nloptr:"
library("nloptr")
is.installed('nloptr')
"rugarch:"
library("rugarch")
is.installed('rugarch')
}
library("rugarch")
#print("RUGARCH IS INSTALLED:")
is.installed('rugarch')
args <- commandArgs(trailingOnly = TRUE)
datatype <- args[1]
size <- args[2]
nr_of_series <- args[3]
differentiation <- args[4]
#print(datatype)
#print(size)
file_name = paste("garch/", datatype, "_", size, ".csv", sep="")
#print(file_name)
#Convert strings to intergers
size_string = size
size = strtoi(size)
nr_of_series = strtoi(nr_of_series)
return_series = read.csv(file_name, header=FALSE)
tmp_serie = return_series[1,]
tmp_time_serie = as.ts(tmp_serie)
windowLength = as.integer(length(tmp_time_serie)*0.10)
forecasts_for_all_series=matrix(nrow=nr_of_series, ncol=size-windowLength, byrow = TRUE) # fill matrix by rows
sigmas_for_all_series =matrix(nrow=nr_of_series, ncol=size-windowLength, byrow = TRUE) # fill matrix by rows
##print(forecasts_for_all_series)
#Judge Order
#final.bic <- Inf
final.order <- c(2,0,2)
#first_serie = return_series[1,]
#first_serie_window = as.ts(first_serie[(1):(windowLength)])
#for (p in 0:5) for (q in 0:5) {
# if ( p == 0 && q == 0) {
# next
# }
#
# arimaFit = tryCatch( arima(first_serie_window, order=c(p, 0, q)),
# error=function( err ) FALSE,
# warning=function( err ) FALSE )
# ##print(arimaFit)
# if( !is.logical( arimaFit ) ) {
# ##print('actually managed to fit')
# ##print(p)
# ##print(q)
# current.bic <- BIC(arimaFit)
# if (current.bic < final.bic) {
# final.bic <- current.bic
# final.order <- c(p, 0, q)
# final.arima <- arima(first_serie_window, order=final.order)
# }
# } else {
# ##print('didnt managed to fit')
# next
# }
#}
print("Final order: ")
print(final.order)
#Start modellling
for (serie_nr in 1:nr_of_series) {
one_serie = return_series[serie_nr,]
one_time_serie = as.ts(one_serie)
if(differentiation == "Once") {
one_time_serie = c(0, diff(one_time_serie[1:length(one_time_serie)]))
}
foreLength = length(one_time_serie) - windowLength - 1
forecasts <- vector(mode="character", length=foreLength)
sigmas <- vector(mode="character", length=foreLength)
truths <- vector(mode="character", length=foreLength)
spec = ugarchspec(
variance.model=list(garchOrder=c(1,1)),
mean.model=list(armaOrder=final.order, include.mean=TRUE),
#mean.model=list(armaOrder=c(final.order[1], final.order[3]), include.mean=TRUE),
distribution.model="sged")
d = 0
max_roll = 100
while (d < foreLength) {
one_serie_window = as.ts(one_time_serie[(1+d):(windowLength+d)])
datapoints_left_in_serie = foreLength - d
nr_of_rolls = min(datapoints_left_in_serie, max_roll)
one_serie_window_future = as.ts(one_time_serie[(1+d):(windowLength+d+nr_of_rolls)])
#one_diffed_serie_window = one_serie_window #diff(one_serie_window)
#spec = ugarchspec(
# variance.model=list(garchOrder=c(1,1)),
# mean.model=list(armaOrder=c(1, 1), include.mean=TRUE),
# #mean.model=list(armaOrder=c(final.order[1], final.order[3]), include.mean=TRUE),
# distribution.model="sged")
model<-ugarchspec(variance.model = list(model = "sGARCH", garchOrder = c(1, 1), variance.targeting=TRUE),
mean.model = list(armaOrder = c(1, 1), include.mean = TRUE),
distribution.model = "norm")
modelfit<-ugarchfit(spec=model,data=one_serie_window, solver='hybrid')
#print(coef(modelfit))
if(is.null(coef(modelfit))){
print("FAILED to CONVERGE")
#Handles exception naively
forecasts[d+1] = 0
sigmas[d+1] = 1
d = d + 1
}
else{
spec = getspec(modelfit);
setfixed(spec) <- as.list(coef(modelfit))
fore = ugarchforecast(spec, n.ahead=1, n.roll = nr_of_rolls, data = one_serie_window_future, out.sample = nr_of_rolls)
for (index in 0:nr_of_rolls) {
forecasts[d+index+1] = fitted(fore)[index+1]
sigmas[d+index+1] = sigma(fore)[index+1]
}
d = d + nr_of_rolls
}
}
forecasts_for_all_series[serie_nr,] = forecasts
sigmas_for_all_series[serie_nr,] = sigmas
cat(".")
}
forecasts_for_all_series <- mapply(forecasts_for_all_series, FUN=as.numeric)
sigmas_for_all_series <- mapply(sigmas_for_all_series, FUN=as.numeric)
forecasts_for_all_series <- matrix(data=forecasts_for_all_series, ncol=foreLength+1, nrow=nr_of_series)
sigmas_for_all_series <- matrix(data=sigmas_for_all_series, ncol=foreLength+1, nrow=nr_of_series)
file_name_mean = paste("forecasts_mean_", datatype, "_", size, ".csv", sep="")
write.table(forecasts_for_all_series, row.names=FALSE, col.names=FALSE, file = file_name_mean)
file_name_var = paste("forecasts_variance_", datatype, "_", size, ".csv", sep="")
write.table(sigmas_for_all_series, row.names=FALSE, col.names=FALSE, file = file_name_var)
print("End of R Program")