Demand planning Forecast errors and KPI to store results of error

[former_member229994]

I have following requirement: keep traceabilitity of hypothesis and forecast error results of former mass process .

In comparison tool, we can compare the result of last interactive forecast run but our requirement is to store those result for KPI purpose.

Is it possible without extra development ? ( built not yet done, we are in design phase)

Kinds regards,

Christine

[Former Member · Accepted solution]

Hi Christine,

There is Standard Error analysis functionality available for forecast accuracy reporting.

Please refer to the URL below:

http://help.sap.com/saphelp_scm70/helpdata/en/6f/8812ce03ba4f0ca09ea31a1a3e1f0c/frameset.htm

Please check if these are sufficient for your business needs.

If we want to extarct standard statistical forecast errors( like MAD, MAPE) calculated in the interactive or batch runs of statistical forecasting , then, to my understanding, development is required. A short detail is given below:

We need to develop macros for calculation of Univariate forecast erros : MAD and MAPE. New planning book and data views may be required to becreated Some extra Key figures also will be needed ,used for storing the values.

The macros runs can be run through batch jobs.These calculated forecast errors of MAD and MAPE then can be compared with the system calculated MAD and MAPE during interactive / batch runs.

The error data can be extracted and stored in the infocubes.

Please check and confirm.

Regards

Datta

[Former Member]

Christine,

If I am understanding correctly, you need to keep track of the forecast errors for multiple runs. You can accomplish this using the "Forecast Comparison" functionality. After you execute the forecast its stored as a version with all the error parameters. I guess upto 10 versions are allowed.

After you enter the univariate forecasting from the planning Book you have the buttin of the "Forecast Comparison" on top left toolbar. There, you can check the errors of different runs, the parameters and the changes. Its a useful functionality for a business which experiences forecast fluctuations or a new product launch. The comparison can help in identifying the best model.

Let me know if this jelps.

Cheers!

Abhi

[former_member229994]

No ,

I know that we can use comparison tool in interactive mode. The requirement is not to get the error interactivly but , in mass forecasting process, to use the result of macro calculating error ( either for alert or to run again for example the best model with strategy 56 ) , to store it in a keyfigure to be able, in BW, to have KPI based on this result.

-> we'll have probably have to research the "internal" (?) macro used for alert or for strategy 56, to trigger alert or other action and re use this macro to polulate a keyfigure "ZALERT" to store result of last mass process run.

Kind regards,

Christine