Forecasting 12 month rolling totals

[PowerBI_77]

Hi everyone.  A Power BI newbie here!  

I'm trying to use Power BI to provide me with a forecast for some 12 month rolling totals.  So for example, the figure for January 2008 would be the sum of figures from February 2007 up to and including January 2008.  It's my understanding that having the data arranged in 12 month rolling totals will smooth out any seasonality impact. 

Here is the same 12 month rolling data on two charts with different forecasting seasonality values. As you can see, the forecast is very different depending on the setting used. 

I'm not sure what setting I should use on the SEASONALITY field on the forecast options?  Does anyone have any idea what value I should use in here please?

Regards,

Lee

[Greg_Deckler]

My understanding for the Seasonality field is that it needs to be set to the period of your seasonality. So, if your period is quarterly, then you should use 4, for example. 

Here are some links that you might find helpful:

https://community.powerbi.com/t5/Community-Blog/Correlation-Seasonality-and-Forecasting-with-Power-B...

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

https://community.powerbi.com/t5/Desktop/Forecast-Functionality-in-PowerBI/td-p/124766

[PowerBI_77]

Thanks for your reply Greg - much appreciated.  It's just the fact that I'm using de-seasonalised data which makes me uncertain about the seasonality setting.  If I plot the data based on monthly totals, I set the seasonality setting to 12 (and, as you say below, if I'm using quarterly data I would set it to 4) which is straight forward.  But when using rolling 12 month data I'm not sure what period my seasonality would be?  I tried applying a value of zero (thinking because my data was de-seasonlised) but Power BI doesn't accept a zero value in that field.

Regards,

Lee