I am trying to add an external regressor xreg into the hts package, however I am getting an error regarding the number of rows (140) despite my external variable has the same number. I have checked other answers but mine is more simple:
Here is the reproducible example
library(hts)
abc <- matrix(sample(1:100, 32*140, replace=TRUE), ncol=32)
colnames(abc) <- c(
paste0("A0",1:5),
paste0("B0",1:9),"B10",
paste0("C0",1:8),
paste0("D0",1:5),
paste0("E0",1:4)
)
abc <- ts(abc, start=2019, frequency=365.25/7)
x <- hts(abc, characters = c(1,2))
data <- window(x, start = 2019.000, end = 2021.166)
test <- window(x, start = 2021.185)
x2 <- runif(n = 140, min = 1, max = 10) #External regressor with the same size
fcastsxreg <- forecast( data, h = 2, method = "comb", algorithms = "lu", fmethod = "arima", weights=, "wls", nonnegative=TRUE, xreg=x2)
accuracy(fcastsxreg, test, levels = 1)
The error message is about the mismatch between de size of abc matrix an x2 vector despite both have 140 rows
Error in model.frame.default(formula = x ~ xregg, drop.unused.levels = TRUE) :
variable lengths differ (found for 'xregg')
In addition: Warning message:
In !is.na(x) & !is.na(rowSums(xregg)) :
longer object length is not a multiple of shorter object length
Thank you
Your training data data has 114 observations for each series in the hierarchy. Your regressor has 140 observations. So there is a difference in length as the error states.
You also need to provide both an xreg argument for the training period, and a newxreg argument for the forecast period.
Another small problem is that there was one observation between your training and test data which was probably unintentional.
Here is a modification of your code that works.
library(hts)
abc <- matrix(sample(1:100, 32 * 140, replace = TRUE), ncol = 32)
colnames(abc) <- c(
paste0("A0", 1:5),
paste0("B0", 1:9), "B10",
paste0("C0", 1:8),
paste0("D0", 1:5),
paste0("E0", 1:4)
)
abc <- ts(abc, start = 2019, frequency = 365.25/7)
x <- hts(abc, characters = c(1, 2))
data <- window(x, end = 2021.166)
test <- window(x, start = 2021.167)
# External regressor with the same size as training and test data combined
x2 <- ts(runif(n = 140, min = 1, max = 10), start = 2019, frequency = 365.25/7)
fcastsxreg <- forecast(data,
fmethod = "arima", nonnegative = TRUE,
xreg = window(x2, end = 2021.166), newxreg = window(x2, start = 2021.167)
)
accuracy(fcastsxreg, test, levels = 1)
#> A B C D E
#> ME -9.278558 11.0833938 -4.7985252 5.8634578 -6.5853672
#> RMSE 58.741525 84.4354712 92.0376431 66.1268442 55.6521141
#> MAE 48.428461 69.6798318 78.7112730 55.7808292 45.1403745
#> MAPE 21.854145 13.4673244 20.5752506 23.0548665 31.0934899
#> MPE -9.642284 -0.5202464 -6.0976324 -5.0032980 -15.7431146
#> MASE 0.638436 0.5647993 0.7868589 0.7730021 0.5935744
Created on 2022-01-26 by the reprex package (v2.0.1)