I'm currently testing out the forecasting option within power BI. I seem to have it working but I'm questioning the veracity of the results. I have daily call data going back over 4 years and have set the seasonality to 365 (which I believe is correct as I have call data per day)? Anyway the forecast is showing a significant drop off in expected calls over the next 12 months which I don't believe is right. The skew appears to be too great. I have included my forecast selections for your viewing pleasure
Any comments would be greatly appreciated.
You might want to check your figures with a different forecasting method:
What happens if you do not put in 365? My understanding is that is an overide but it should automatically detect the seasonality. But, I believe that if you have daily data that 365 should be correct. I would also refer you to these articles:
Proud to be a Datanaut!
Thanks for responding. I removed the 365 value from the "Seasonality" text box and left it on "Auto" as you suggested. The results look a lot better and more inline with the previous years. Does this mean that PowerBI is utilising a Seasonality value different to 365? If so is there a way to see what it is using? I'd like to know more about that for my own understanding of Seasonality here.