Running Statistical Forecast Jobs with complex selection at the aggregate

[Former Member]

Hi I am looking for some answers regarding running statistical forecasting profiles in the mass processing jobs, using transaction /SAPAPO/MC8D - Create Demand Planning in the Background to create the batch jobs.

Currently in Demand Planning you can create a complex selection ID with different characteristics and materials that span accross different hierarchies and load the entire selection into the DP grid and run a statistical forecast model for the aggregate.

However when you set up jobs with /SAPAPO/MC8D you have to select an aggregation level so you can not run it in background at the selection ID aggregate. Is there anyway around doing this other than interactively?

[Former Member]

Sure. Our Characteristic combinations consist of the following characteristics.

u2022 Material

u2022 Country

u2022 Sales Org

u2022 Forecast Group

u2022 Forecast RDC

We also have navigational attributes attached to the Material characteristic:

u2022 Business Unit

u2022 Category

u2022 Marketing Product

u2022 Price Group

We can use any of those in a selection ID to pull up data and load into the demand planning grid in /SAPAPO/SDP94.

So for example if our selection includes:

Show Marketing Product

that meet the followng conditions

Planning Version: DP1

Country: US

Marketing Product: 100002, 100004

In the "Selected Objects" window, the two Marketing Products will show up. You can load both of these at the same time in the grid and run a statistical forecast against them.

If I set up a DP Background job with /SAPAPO/MC8D, you have to select an aggregation level after choosing the activity (which includes the forecast profile to be run) and selection ID to run the job for. The system will give you a hard error to select a aggregation level. The levels include all of the characteristics or navigational attributes. I am forced to choose one so I pick Marketing Product but then the job will run the statistical model at each individual Marketing product level. So essentially to forecast model executions, 1 for 100002 and 1 for 100004. This produces a different result.

The business is asking for the ability to run the background jobs at the selection ID aggregate level that is available in interactive demand planning.

Thanks for your help!

[Former Member]

Hi Stacey.

For your example, if you selected country as your aggregation level against the selection then you should get the result you're after.

Hope this helps, M.

[Former Member]

I apologize for a miss type but there is another country value in the selection. It should be US and CA.

Here are the 2 marketing products with the list of the materials that are included and the country that they belong to. Is there any way to run a statistical forecast at the aggregate of something like this?? Your help is greatly appreciated!!

Marketing Product Materials US CA

100002 23234 x

3234 x

3235 x

33235 x

33237 x

44542 x

53237 x

603731 x

63234 x

73235 x

100004 74519 x

[Former Member]

Hi Stacey.

If I understand your data (and I'm not 100% clear), I would say that it couldn'e be done as is. A possible way to do it would be to create a new navigational attribute with some logic that links together the values of one of the characteristics. (eg. 'Forecast Aggregation Attribute') and assign values accordingly. This could be used in the job definition.

Beware however, forecasting at the level of navigational attribute introduces some limitations for lifecycle planning.

Hope this helps, M

[Former Member]

Thank you very much for your answer. I was thinking that the only way to do it was to create a navigational attribute of some kind but I am still not sure this would get what we need. I was hoping there would be some config setting or enhancement note I could implement that would provide this functionality. I will work with the BI development team to see if this is something we can accomplish.

Can you give more specifics on what the limitations and impact forecasting at the navigational attribute level has on life cycle planning? This is interesting and something I have not previously thought about but may need to introduce to the business team.

Thank you again for your help!

[Former Member]

HI Stacey.

As far as the limitations on lifecycle planning go.

If you are aggregating above characteristic level when running forecasting and you are assigning lifecycle components at the characteristic level, you have to use 'Aggregated Lifecycle Planning' (flag in the lifecycle settings for the planning area). This does work, however, you cannot use this in conjunction with 'Number of Days in Period' to drive the calculation of corrected history and forecast. At some clients, if aggregated lifecycle planning has been required, have had to calculate corrected history and forecast in planning book macros outside of the forecasting run.

Hope this helps, M.

[Former Member]

I cannot find this flag in the lifecycle settings for the planning area. I looked in //MSDP_ADMIN and also the planning book configuration. We do have the Lifecycle planning check box checked in our master forecast profiles but I cannot find this anywhere else.

Thanks again for your help.

[Former Member]

Hi Stacey.

The flag is set on the lifecycle planning settings for the planning area. I don't have a system to hand to tell you the transaction, but in the menu is Demand Planning->Environment->Lifecycle Planning. Select Basic Settings.

Help file link gives some extra info you may be interested in:

LInk: [http://help.sap.com/saphelp_scm2007/helpdata/en/58/a0d962d7f2154094fc57cb8a564e85/content.htm|http://help.sap.com/saphelp_scm2007/helpdata/en/58/a0d962d7f2154094fc57cb8a564e85/content.htm]

Hope this helps, M.

Edited by: percx on Sep 28, 2008 7:49 PM

[Former Member]

can you be more specific regarding the complex selection Id as we can select the aggregation level in /SAPAPO/MC8D as well