Click here to monitor SSC
SQLServerCentral is supported by Redgate
 
Log in  ::  Register  ::  Not logged in
 
 
 
        
Home       Members    Calendar    Who's On


Add to briefcase «««123

Forecasting with SQL Expand / Collapse
Author
Message
Posted Wednesday, February 1, 2012 3:11 PM
Grasshopper

GrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopper

Group: General Forum Members
Last Login: Tuesday, November 13, 2012 2:21 PM
Points: 10, Visits: 151
Hi Mark,

Nice article on forecasting using T-SQL! I know it's been a while since you wrote this article but wanted to know if you think this technique would work with a scenario I am faced with on a project.

My scenario is a table of risks with a value for severity, probability and a criticality index value. I would like to predict or forecast the number of risks most likely on the fly so not stored in a table for a charting solution. Basically, I would chart the actual values and then based upon a choosen date display the forecast values to the right and the actual values to the left of the choosen date.

As of now I am rolling up the number of risks by a category or risk type for a monthly rollup which works fine. I'd like to take this data and then do a forecast for future months like you do in this article for say 6 to 12 months ahead from a choosen (parameter) now date.

So in your article you use products and in my scenario I have risks. Do you think I could use this technique to do the forecasts? The forecasts would only have to project an integer or count value based on the historical number of risks in previous months of a project.


Thanks in advance!





Chad E. Downey, CDMP - Certified Data Management Professional
Consultant - SQL Server, SSIS, SSRS, SSAS
Post #1245360
Posted Wednesday, February 1, 2012 8:08 PM
Grasshopper

GrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopper

Group: General Forum Members
Last Login: Thursday, July 9, 2015 1:44 PM
Points: 17, Visits: 581
Yes. That should work fine for you to give a directional insight into your data. Products, customers or risk types are simply discrete values by which to partition the forecast to make it more useful. Any type of discrete data can be plugged in.

Good luck and have fun exploring forecasting!
Mark
Post #1245437
Posted Tuesday, October 30, 2012 6:51 AM


SSC Veteran

SSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC VeteranSSC Veteran

Group: General Forum Members
Last Login: Tuesday, April 26, 2016 8:48 PM
Points: 289, Visits: 1,320
Its very useful for the beginner. If the same sample, if available using SSAS would be great...

Many Thanks!
S.saravanan
“I am a slow walker, but I never walk backwards-
Abraham Lincoln”
Post #1378742
Posted Monday, March 21, 2016 6:22 PM
Forum Newbie

Forum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum Newbie

Group: General Forum Members
Last Login: Monday, March 21, 2016 7:07 PM
Points: 1, Visits: 12
Mark Lovely work and making my life much easier for a painful problem of how to cast time series data forward.

On addition i would to understand is how to make ll the NULL values come through as blank so I can automatically process them into a charting web service.
There seems to be multiple places as I have tried to zero them out but i would prefer blanks so the charting does not plot them at all.

I will go hunting for more tidbits you have uploaded to broaden my knowledge.

Thanks a lot
Post #1771190
« Prev Topic | Next Topic »

Add to briefcase «««123

Permissions Expand / Collapse