0

I've got data of which I think it has a seasonality. My data has a peak in july/august and one in december. I have only data of 2014 and 2015, but in both of the cases this is happening. (See my graph)

I tried to confirm this with autocorrelation, but this method only gives the so-called "White noise", probably because the seasons are 'irregular' (not one peak per year or something, it's not yearly, not quarterly etc.)

I need to forecast over this data, but I'm not sure if I should pick a model that uses seasonality or one that doesn't use seasonality.

Data

enter image description here

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Grafit
  • 155
  • 8
  • You can find this thread about detecting seasonality helpful: https://stats.stackexchange.com/questions/16117/what-method-can-be-used-to-detect-seasonality-in-data – Tim Feb 22 '16 at 13:36
  • 1
    There's nothing that can be done with seasonality if you have data for only one cycle. Unless, of course you have a very good theory behind the data. – Aksakal Feb 22 '16 at 14:31
  • @Aksakal I've got two cycles, 2014 and 2015, right? – Grafit Feb 22 '16 at 15:34
  • Ok, two cycles is better than one :) Just put the dummy variable on "summer" months. – Aksakal Feb 22 '16 at 15:50
  • @Aksakal Even when my autocorrelation doesn't show seasonality? Do you maybe have an example of a model with a dummy variable? Or could I just add it with a certain parameter to any model? – Grafit Feb 23 '16 at 08:13
  • @Tim I see now I shouldn't use autocorrelation with irregular seasonality, are there other methods except autocorrelation I can use to determine seasonality? I have only 24 data points of 24 months so I don't think the power plot of the link you posted is usable for me – Grafit Feb 23 '16 at 15:22

0 Answers0