Calculation of the Standard Deviation and MAD 

Use

The system calculates the standard deviation and the mean absolute deviation (MAD). This standard deviation is also used by other planning services in Service Parts Planning, such as inventory planning.

Features

The calculation of the standard deviation and the MAD is part of the forecast model itself in the following forecast models:

●      intermittent forecast model

●      seasonal trend model with FPG

For all other models the system uses the historical forecast to calculate the standard deviation and the MAD.

The system calculates the standard deviation as follows:

The system calculates the MAD as follows:

...

       1.      The system initializes the first three of the historical periods you defined for calculating the MAD

The system proceeds as follows:

You can enter weighting factors in Customizing under Initialise Base Values. For more information, see the Implementation Guide (IMG) for Advanced Planning and Optimization under Supply Chain Planning → Service Parts Planning (SPP) → Forecasting.

       2.      The system calculates the MAD according to the following formula:

i represents the current period.

You can specify alpha from the SAP Easy Access menu under Advanced Planning and Optimization → Service Parts Planning (SPP) → Planning → Forecast → Forecast Profile on the Model Parameter tab page in the Alpha Factor field. In the DMA model you specify alpha on the Model Parameter tab page in the MAD Smoothing Factor for Demand in DMA Model field.

i-RLT represents the period, in which the forecast for period i was created. You can specify this value in the forecast profile on the General tab page in the Offset Forecast Creation field. 

n is the number of periods that you defined in the forecast profile on the Model Parameter tab page.

The following table gives you an overview of which parameters you must define for which forecast model:

Forecast Model	Parameter
First-order exponential smoothing	Periods StdDev./MAD for First-Order Exponential Smoothing
Second-order exponential smoothing	Periods StdDev./MAD for Second-Order Exponential Smoothing
Moving average	Periods StdDev./MAD for Moving Average
Linear regression model	Periods StdDev./MAD for Linear Regression
Seasonal trend model	Periods StdDev./MAD for Seasonal Trend Model
Seasonal trend model with fixed period groupings (FPG)	Periods StdDev./MAD for Seasonal Trend Model with FPG
Intermittent forecast model	Periods StdDev./MAD for Intermittent Forecast Model
Dynamic moving average	Periods StdDev./MAD for Dynamic Moving Average
Declining demand forecast (DDF)	Periods StdDev./MAD for Declining Demand Forecast