r - for loop only stores last value -

i wrote loop in r , try store computed values in matrix. unfortunately, saves last value. browsed internet , found out, problem can solved indexing. however, , still doesn't work. see mistake?

data set:

require(stats)  data <- ts.union(a=arima.sim(model=list(ar=c(.9,-.2)), n=144),               b=arima.sim(model=list(ar=c(.6, -.3)), n=144),               c=arima.sim(model=list(ar=c(-.2,-.6)), n=144),               d=arima.sim(model=list(ar=c(-.1,-.6)), n=144),               e=arima.sim(model=list(ar=c(.2,-.6)), n=144),               f=arima.sim(model=list(ar=c(.2,.7)), n=144),               g=arima.sim(model=list(ar=c(.3,.2)), n=144),               h=arima.sim(model=list(ar=c(-.4,.3)), n=144),               i=arima.sim(model=list(ar=c(.1,-.7)), n=144),               j=arima.sim(model=list(ar=c(.8,.1)), n=144)) data <- ts(data, start=c(2007, 2), frequency=12) 

loop:

require(vars)  for(i in 1:12){   # compute factors, window 5.5 years wide   factors <- ts(prcomp(window(data, start=c(2007, (i+1)), end=c(2012, i+6)), center=t, scale=t)$x[,1:5],            start=c(2007,(i+1)), frequency=12)   # estimate var model   model <- var(window(data, start=c(2007, (i+1)), end=c(2012, i+6)),           exogen=factors, type="const")   # forecast factors   factor.fcst <- sapply(factors, function(x) predict(auto.arima(x, stationary=t, seasonal=f), n.ahead=12))[1,]   factor.fcst <- cbind(factor.fcst$pc1, factor.fcst$pc2, factor.fcst$pc3, factor.fcst$pc4, factor.fcst$pc5)   colnames(factor.fcst) <- colnames(factors)    # forecast model   a_fcst <- ts(predict(model, dumvar=factor.fcst, n.ahead=12, ci=0.95)$fcst$a[,1],               start=c(2012, (i+7)), frequency=12)   # compute rmse   rmse <- matrix(ncol=1, nrow=12)   rmse[i,] <- sqrt(mean((window(data[,1], start=c(2012, (i+7)),                            end=c(2012, (i+18))) - a_fcst)^2))   print(rmse) } 

thanks help!

here's simpler example:

  d1 <- seq(10)   ans1 <- vector(length=length(d1), mode="double")   (i in seq.int(length(d1))){     ans1[i] <- sqrt(d1[i])   } 

now, if take second line , put inside loop, last value returned i.e. value of ans1 reset each time loop run.

this sort of code far 'best practice'. said, loops can useful first step towards vectorizing function, can easier read , (rarely) can faster.