Skip to main content
7 events
when toggle format what by license comment
Mar 16, 2021 at 13:34 comment added Chesso Here is the answer to my previous question: otexts.com/fpp2/seasonal-arima.html
Mar 16, 2021 at 13:29 comment added Richard Hardy I think the answer should mention the glaring problem of overdifferencing, i.e. differencing data that does not contain unit roots but contains deterministic trends. You have addressed differencing but not in this way.
Mar 16, 2021 at 13:15 comment added Chesso I have completed this task, however It produces a notation I have not seen before. What is meant by ARIMA(0,1,2)(0,1,1)[12]? If this should be asked in the R section of stackoverflow, please let me know.
Mar 16, 2021 at 10:43 comment added Stephan Kolassa Significance only tells you about in-sample fit. If that is all you want, go ahead, but usually people want to forecast their time series, and then in-sample fit gets misleading very quickly. I would really recommend relying on auto.arima() as to differencing - it's very hard to beat.
Mar 16, 2021 at 10:36 comment added Chesso Understood. I have fitted my initial plot, and have found a cubic model to be highly significant. So in the case of differencing, should I not be using a third order difference? Use a lower order difference? Not use differencing at all?
Mar 16, 2021 at 10:34 vote accept Chesso
Mar 16, 2021 at 7:37 history answered Stephan Kolassa CC BY-SA 4.0