I'm using the TBATS model of the forecast package with Google Analytics data, to forecast web trafic containing multiseasonal effects (msts).

I have two year of daily data that contains two columns like these :

date           sessions
2015-01-01     2667
2015-01-02     3542
2015-01-03     2383
2015-01-04     2772
2015-01-05     7797
2015-01-06     7599

I made one forecast for all my data and another forecast with data containing only a segment of users.

In the first case (all data), I obtained this result :

enter image description here

And in the second case (segment data), I obtained this :

enter image description here

My questions :

  1. How to interpret the slope component? I get it in the first case but not in the second. Is there a reason for that?
  2. What could do that the Lo/Hi points are so much wider in the second case?

In your first example two years of annual data was used with no apparent effects due to omitted variables like price , promotion etc or level shifts . The two years of forecasts are visually identical with presumptive symmetric forecasting confidence limits totally ignoring any unusual activity in the future. Your second example based upon only 1 year of data has less predictability (fitting statistics ... smaller r-square ...larger error variance) thus wider limits would be expected as no seasonal component was identifiable due to only 1 year of data.

  • $\begingroup$ if you post your data I will take a look at it and provide you with a model that will convert your data to information and perhaps some knowledge/insight.. Please specify the starting date and the country of origin. $\endgroup$ – IrishStat Apr 9 '17 at 1:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.