cancel
Showing results for 
Search instead for 
Did you mean: 

Demand statistical model request

Former Member
0 Kudos

Good day,

I´m a worker from Alicorp S.A.A., a peruavian company that use the SAP SCM 5.00 Release 01J_SCM500 .Our company always use the u201CSCM-APOu201D system to get the statistical forecast, but we don´t feel very comfortable seen a number without knowing how the system is getting the result, so I´m asking if you could tell me the formula of the following statistical models (I don´t know the exact name in English), that are the one that we use.

M1: Constant model with fist order alfa adaptation (number 12)

M2: Holt process (number 21)

M3: seasonal lineal regression (number 35)

M4: Forecast with seasonal trend model (number 40)

M5: Selection with process 2 model (number 56)

And the formula for the errors in the five models.

Also, could you tell me how the months are taken when you work with 24, 12 and 6 months, the system go back that period and consider all the months until the last month or only take a amount of months starting 24 months back in time.

Thank you very much.

Fabio Silgado

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Fabio,

Discussion of all the univariate (time series based) forecast strategies, along with the underlying formulas can be found in standard SAP online help

http://help.sap.com/saphelp_scm70/helpdata/EN/46/a4dee34d540487e10000000a1553f6/frameset.htm

Please don't ask me to explain the math behind these models, I am definitely not a statistics expert. SAP has merely adopted the most commonly used mathematical models in APO. To ease your doubt, SAP also offers an automated way to select the 'best' forecast, based on statistical data fit, similar to most forecasting software.

Generic APO Forecasting info

http://help.sap.com/saphelp_scm2007/helpdata/en/33/437a37b70cf273e10000009b38f8cf/frameset.htm

When you design your forecasting strategy, YOU determine which periods to analyze, from which to derive the forecast. For most businesses, when they refer to a '24 month' forecast, they are referring to the most recent 24 months worth of data.

.

Rgds,

DB49

Answers (0)