Forum Discussion
Flat forecast line when using Auto Seasonality
Hi All,
i am working on a project to forecast the next ~5 months sales revenue. I conducted ACF, PACF and decomposition on my data set and found that there was no seasonality.
I have added the data into power bi desktop, aggregated the data into weeks (i cannot use days as each day does not have a value - but each week does so i used week ending), i then created a line chart and turned the forecast feature on. Given i proved there was no seasonality by using python i left the seasonality as "auto" - power bi is meant to detect any seasonality anyway and choose the right forecast model (seasonal vs non seasonal) - but for some reason each time i try and no matter what i change on the units or forecast length the forecast line is static and flat giving the same predicted value for each week. Any ideas why this may be and if there is a way to fix it? I cannot add a number to seasonality as i proved there was no seasonality in my data - my assumption here was Power BI would work that out for itself and i leave seasonality as auto and it would give me a forecast but that isn't happening.
1 Reply
- lbendlin
Super User
it means that Power BI has not discovered any discernible pattern in your data (with or without seasonality). Too much jitter. It basically gives you back the average.