I have used the
auto.arima function with my data (n=156, frequency = 52) The residuals look like this:
The corresponding Q-Q plot shows the following progression:
Due to the one year's lead time and the small amount of data, the
qline() is not lying on the points correctly.
However, if I take away the first year of residuals and look at the last two years only (n=104), the
qqplot() looks like this:
In my opinion, this corresponds more to the "target output". My question would be: Can I simply leave out the one year in the plot? And if not, how should I interpret the actual Q-QPlot and what exactly does it say?
Thank you very much for your help.