MS Time Series: Absurd predicted number when stdev = 1

Question

I have noticed that for the prediction of MS Time Series, the number is usually absurdly large when the standard deviation is 1.  I always get a bunch of them. Does anyone have any clue about what this means?

Answer 1

In most cases, we stop predicting with the time series algorithm when we detect that predictions are not stable, so you'll see NULL values for future steps beyond a certain number. You might be hitting a corner case where this doesn't happen.

Answer 2

If you can share some data that causes this, I'd be glad to take a look.

--Alex--

http://www.research.microsoft.com/users/alexeib/

http://www.sqlserverdatamining.com/DMCommunity/

 

Answer 3

Thanks! Please allow me some time to put the data  back together and send you.