Skip to Content
0

Sum/average the past 6 periods in IBP

Dec 01, 2017 at 09:14 PM

226

avatar image
Former Member

Hi,

We have the business requirement to create a bias tracking signal report that has a rolling total forecast variance (fcst-actuals) sum over the past 6 periods (including the current period) and the sum of the absolute value of the past 6 periods. Then calculate the absolute sum of the past 6 periods divided by 6 to get a MAD. The final result is the total forecast variance (without absolute) / MAD. We have referenced the link below for this situation however, we believe it will probably require L code script for us to go that route with periodid transformation for calculating the 6 past periods. Please advise if anyone has found a way to calculate this without using L code. Maybe local member if there's example for this situation or if the periodid transformation can work would be possibilities.

https://answers.sap.com/questions/132108/past-6-weeks-average-sap-ibp.html

Thanks,

Melanie

10 |10000 characters needed characters left characters exceeded
* Please Login or Register to Answer, Follow or Comment.

1 Answer

Best Answer
Matheus Korndoerfer
Dec 04, 2017 at 12:36 PM
0

Hello Sanjeev,

Have you also checked Note 2240170? It has a PDF SAP Integrated Business Planning Advanced Configuration - Rolling Sum of Last Three Periods (Attribute Transformation).pdf attached that explains how to configure a Key Figure with sum from the last three periods (starting from the current period) on a rolling basis. You may just change the SUM by the AVG, in this case.

Best regards,
Matheus

Show 4 Share
10 |10000 characters needed characters left characters exceeded
Former Member

Hi Matheus,

Yes, this is the example I'm referring to but I'm wondering if there is possibly an easier way to do this, maybe using local members?

In the example you refer to, we only need monthly so would we only need to do transformations for PERIODID0? Also, it says to configure PERIODID0 +1, PERIODID0 + 2, etc. Is that supposed to be PERIODID0 - 1, PERIODID0 - 2, etc. because we are looking backwards from the current month? In this case, the base planning level is at the MTH-PROD-LOC-CUST level, do we need to make 5 copies of this level in order to do this configuration? It looks like in the example, it copies the base planning level twice.

Thanks,

Melanie

0

Hi Melanie,

Yes, Local Members should work too by using a simple average function. You can refer to our SAP Help Portal EPM Report Editor explaining how to create Local Members.

Regarding the usage of PERIODIDn, it really depends on the configuration of your Time Profile. For example, if you have Day, Weeks, Tech Weeks, Month and Year, this is how the PERIODIDn are defined:
You may also refer to the Model Configuration Guide - PERIODID and PERIODID(n) Attributes in Time Profile Levels.

  • PERIODID0: Daily periods
  • PERIODID1: Yearly periods
  • PERIODID2: Monthly periods
  • PERIODID3: Weekly periods
  • PERIODID4: Technical weekly periods

For the PERIODID0 +1 and +2, you should follow the instructions of the Note. In case it calculates in future periods, you may try using PERIODIDn -1 and -2. And as you need 5 backwards periods, you should indeed configure 5 Planning Levels at the MTH-PROD-LOC-CUST.

Best regards,
Matheus

0
Former Member

Thank you Matheus! I was able to get the bias tracking signal to work with PERIODID transformation with just a few differences from the pdf including the buckets, periodidn, planning levels, and some of the stored values.


Thanks,


Melanie

1

Hi Melanie,

Nice! Thank you for the confirmation!

If you want, feel free to share some important configuration changes you had to made in case other colleagues face a similar issue.

Best regards,
Matheus

0