Forecasting using SAS Forecast Server, SAS/ETS, and more

Changes to the Data Generating Process

Reply
Contributor
Posts: 62

Changes to the Data Generating Process

Hi SAS Friends,

I apologize for the slightly out of place question but I figured the great resources here could help Smiley Happy

My issue is that I have time-series data that trends nicely expect for this last quarter where the data generating process was changed. now some of the estimates don't trend. are there any statistical techniques i can apply to smooth out the data? i only have this last data point using the new data-generating process.

Thanks for your help,

also if you'd like to recommend a statistical/econometric forum where you think this question would be better served i'm all ears.

Respected Advisor
Posts: 2,655

Re: Changes to the Data Generating Process

With only a single point, I don't see how you can do much of anything.  With a couple of points, a Kalman filter might help, but it would still be greatly influenced by the historical values.

Can you apply the new data generating process to old source data in any way, such that you could detect a level change or a multiplier?

Not much help, I'm afraid.

Steve Denham

Ask a Question
Discussion stats
  • 1 reply
  • 400 views
  • 0 likes
  • 2 in conversation