plotting the m-step ahead fitted values in the VAR models: Principles of forecasting

[nickcox896]

Suppose we have this VAR model:

library(fpp3)
fit <- us_change |>   model(VAR(vars(Consumption, Income)))

I compute, say, the 4-step ahead fitted values

fits12<-fitted(fit, h=4)

Now I am plotting both the real and the fitted values along with,say, the h=10 step ahead out-of sample forecasts

fit|>
 forecast(h = 10) |>
  autoplot(us_change) +
  autolayer(fits12, .fitted, col = "#D55E00")  

but my code does not work very well as I get the plots of the out of sample forecasts and of the real time series but
I do not get the plots of the fitted values.

[robjhyndman]

Please take the time to do some basic checking before asking a question. The first step when you have a problem with a graph is to look at the data you are trying to plot. Just type fits12 in the console, and you will see that it contains only missing values. So the question is not why don't you get the plots, but why aren't you computing the fitted values?

In this case, multi-step fitted values have not been implemented yet for VAR models. I'll update the function so you get a warning.

[nickcox896]

Thank you very much. I noticed the Nas and in my data set but I thought that I was implementing in a wrong way the fitted function. So i thought that the mistake was coming from me. Now it is clear. Thanks a lot