cancel
Showing results for 
Search instead for 
Did you mean: 

Measurement of Forecast errors in APO DP

Former Member
0 Kudos

I am using APO DP version 5.

It is likely that I will need to use the BADI /SAPAPO/SDP_FCSTERR for a user specific measure of forecast error.

Does anyone have any experience of using this BADI?

For example if I want to have a user specific MAPE based on [absolute(actual minus forecast) / (forecast)] rather than the standard MAPE based on [absolute(actual minus forecast) / (actual)], could I use the BADI for this?

Thanks,

Bob Austin

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Yes Bob, you have found the right BADI.

Only thing is, this BADI won’t compute MAPE errors, but will calculate "Customer-Spec. Forecast Errors", using the formula you provide.

This BADI will provide you an internal table called IT_FCSTVIEW which shows the history and forecast specific keyfigures. You can do your manipulations as you wish on these values to calculate EV_FCST_ERRPR, shown as "Customer-Spec. Forecast Errors" in the output.

Rgds,

Sarah.

Former Member
0 Kudos

Sarah,

Many thanks for your prompt and helpful reply.

As is often the case, some 'follow up' questions!...

1. I need to be able to report on the calculated forecast error; can the BADI be used to populate a key figure with the error data so that reporting can be against this key figure?

2. I assume the BADI can be used in both interactive and mass processing forecasting?

3. Do you by any chance have any screenshots of how you have used the BADI which you can share with me (email: bob.austin@atosorigin.com)

Regards, Bob.

Former Member
0 Kudos

Bob,

1. This particular BADI can't be used for reporting the errors, unless you code for a Ztable population, but there are ways to do acheive this.

2. Not tested if BADI works for both interactive and mass processing forecasting. Logically it should.

3. Use of BADI is ABAP-wise simple, as it's a math of keyfigures in an internal table, each record representing a time bucket (week/month) same as Pbook. So the internal table would have records = ( history + forecast periods), and only 1 float variable is the output of error calculation formula.

Regards, Sarah.

Former Member
0 Kudos

Hi Sara,

I'm dealing with the same problem, as far as I see, there should be a line into the table which has actual and other with forecast so you can use any mathematical function you want in order to get your ZMAD, ZMAPE, etc.

Our customer wants to use it's own method to error measurement, with the last four/five points, I don't consider it correct however we are wondering how to do it.

Any comment will be useful.

Regards,

Carlos Eduardo Rodríguez C.

Inteligencia de Negocios - APO & BW

Centro de Soluciones de Planeamiento y Estrategia

BELCORP

+51(1) 2113300 Ext. 2884

Tempus rerum imperator

Answers (0)