forecasting

[Anonymous]

Hi All,

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.

Regards

Glenn

[Greg_Deckler]

You might want to check your figures with a different forecasting method:

https://community.powerbi.com/t5/Quick-Measures-Gallery/De-Seasonalized-Correlation-Coefficient/m-p/...

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:

https://powerbi.microsoft.com/en-us/blog/describing-the-forecasting-models-in-power-view/

https://powerbi.microsoft.com/en-us/blog/power-bi-desktop-september-feature-summary/

https://community.powerbi.com/t5/Community-Blog/Predict-your-milestones-with-forecasting-in-Power-BI...

https://docs.microsoft.com/en-us/power-bi/desktop-analytics-pane

https://datamadegood.com/2016/10/02/forecasting-the-future-now-in-power-bi/

[Anonymous]

Hi Greg,

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.

Regards,

Glenn