Send output to:
Browser Blue - Charts White
Browser Black/White
CSV
Data:
4352 4320 4440 4408 4426 4404 4427 4375 4327 4336 4325 4365 4305 4282 4256 4181 4163 4268 4212 4220 4243 4243 4245 4221 4195 4219 4219 4198 4103 4122 4062 4061 4015 3927 3938 3927 3904 3810 3755 3728 3700 3637 3584 3551 3500 3428 3433 3385 3391 3354 3334 3327 3169 3169 3130 3094 3074 3036 3018 2999 2932 2943 2926
Type of Seasonality
additive
additive
multiplicative
Seasonal Period
12
12
1
2
3
4
5
6
7
8
9
10
11
12
Chart options
R Code
par2 <- '12' par1 <- 'additive' par2 <- as.numeric(par2) x <- ts(x,freq=par2) m <- decompose(x,type=par1) m$figure bitmap(file='test1.png') plot(m) dev.off() mylagmax <- length(x)/2 bitmap(file='test2.png') op <- par(mfrow = c(2,2)) acf(as.numeric(x),lag.max = mylagmax,main='Observed') acf(as.numeric(m$trend),na.action=na.pass,lag.max = mylagmax,main='Trend') acf(as.numeric(m$seasonal),na.action=na.pass,lag.max = mylagmax,main='Seasonal') acf(as.numeric(m$random),na.action=na.pass,lag.max = mylagmax,main='Random') par(op) dev.off() bitmap(file='test3.png') op <- par(mfrow = c(2,2)) spectrum(as.numeric(x),main='Observed') spectrum(as.numeric(m$trend[!is.na(m$trend)]),main='Trend') spectrum(as.numeric(m$seasonal[!is.na(m$seasonal)]),main='Seasonal') spectrum(as.numeric(m$random[!is.na(m$random)]),main='Random') par(op) dev.off() bitmap(file='test4.png') op <- par(mfrow = c(2,2)) cpgram(as.numeric(x),main='Observed') cpgram(as.numeric(m$trend[!is.na(m$trend)]),main='Trend') cpgram(as.numeric(m$seasonal[!is.na(m$seasonal)]),main='Seasonal') cpgram(as.numeric(m$random[!is.na(m$random)]),main='Random') par(op) dev.off() load(file='createtable') a<-table.start() a<-table.row.start(a) a<-table.element(a,'Classical Decomposition by Moving Averages',6,TRUE) a<-table.row.end(a) a<-table.row.start(a) a<-table.element(a,'t',header=TRUE) a<-table.element(a,'Observations',header=TRUE) a<-table.element(a,'Fit',header=TRUE) a<-table.element(a,'Trend',header=TRUE) a<-table.element(a,'Seasonal',header=TRUE) a<-table.element(a,'Random',header=TRUE) a<-table.row.end(a) for (i in 1:length(m$trend)) { a<-table.row.start(a) a<-table.element(a,i,header=TRUE) a<-table.element(a,x[i]) if (par1 == 'additive') a<-table.element(a,signif(m$trend[i]+m$seasonal[i],6)) else a<-table.element(a,signif(m$trend[i]*m$seasonal[i],6)) a<-table.element(a,signif(m$trend[i],6)) a<-table.element(a,signif(m$seasonal[i],6)) a<-table.element(a,signif(m$random[i],6)) a<-table.row.end(a) } a<-table.end(a) table.save(a,file='mytable.tab')
Compute
Summary of computational transaction
Raw Input
view raw input (R code)
Raw Output
view raw output of R engine
Computing time
0 seconds
R Server
Big Analytics Cloud Computing Center
Click here to blog (archive) this computation