ARIMA with regressions for Hierarchical data forecast

Question

I am getting this error when trying to use ARIMA with regressions on a gts object:

Error in ...fourier(x, K, length(x) + (1:h)) : K must be not be greater than period/2

Here's a simple reproducible piece of code. What should I set k to? I've tried different values but none seem to work.

library(hts)
y3 <- ts(matrix(rnorm(300),ncol=60,nrow=5))
blnames3 <- paste0(rep(c("CA", "NY"), each = 30), # State
               rep(c("AL", "LA", "CL", "ES"), each = 15), # County
               rep(c("O", "O", "O", "C", "C"), 12), # Industry
               rep(c("p", "q", "r", "p", "q"), 12),  # Sub-industry
               rep(504:507, 15)) # Product
colnames(y3) <- blnames3

gy3 <- gts(y3, characters=list(c(2,2),c(1,1,3)))

i=5
fc <- forecast(gy3, fmethod="arima", seasonal=FALSE, h=6, xreg=fourier(gy3, K=i), newxreg=fourierf(gy3, K=i, h=6))

Answer 1

There are several problems here. First, your time series contain only 5 observations each which is too few to fit any model for forecasting purposes. Second, you don't specify the frequency of the data in the ts() call. Third, you cannot pass a gts object as the first argument to fourier().

Here is some code that works:

library(hts)
y3 <- ts(matrix(rnorm(3000),ncol=60), frequency=12)
blnames3 <- paste0(rep(c("CA", "NY"), each = 30), # State
                   rep(c("AL", "LA", "CL", "ES"), each = 15), # County
                   rep(c("O", "O", "O", "C", "C"), 12), # Industry
                   rep(c("p", "q", "r", "p", "q"), 12),  # Sub-industry
                   rep(504:507, 15)) # Product
colnames(y3) <- blnames3

gy3 <- gts(y3, characters=list(c(2,2),c(1,1,3)))

i <- 5
fc <- forecast(gy3, fmethod="arima", seasonal=FALSE, h=6,
        xreg=fourier(y3[,1], K=i), newxreg=fourierf(y3[,1], K=i, h=6))

Note that the first argument to fourier and fourierf is only used to determine the frequency and length of the Fourier predictors. So just using the first column of y3 is sufficient.