Planning Level: Forecast 

Technical name: 4TPM8107

Characteristics and Key Figures

The planning level Forecast (4TPM8107) contains the key figures Net Revenue ( 0COPANETRV) and Billing Document Quantity in Sales Unit of Measure ( 0INV_QTY).

It also contains the following characteristics:

Characteristics

Characteristic	Description	Selection
0CRMVERSION	CRM Plan Version	CPVERS
0BP_GRP	BP: Group (Hierarchy)
0FISCPER	Fiscal Year/Period	FISCPER03
0FISCVARNT	Fiscal Year Variant	K4
0COUNTRY	Country Key
0CRM_PROD	Product
0DIVISION	Division
0SALES_UNIT	Sales Unit
0CRM_SALORG	CRM Sales Organization
0VERSION	CRM Version
0DISTR_CHAN	Distribution Channel
0VTYPE	Reporting Value Type
0CURRENCY	Currency Key	EUR

Planning Packages

The planning level contains two planning packages:

·        The planning package Forecast: CMP Version (4TPM8187) restricts the version to the value CMP (business plan).

·        The planning package Current FC: CMP & CAF Versions (4TPM817) restricts the version to the values CMP (business plan) and CAF (current forecast).

Planning Layouts

The planning layout Forecast Overview (4TPM8157) displays the current forecasts for the key figure Billing Quantity (0INV_QTY) with reference to the relevant customer groups and products. You cannot make any changes to this planning layout.

Planning Functions

The following planning functions, which are used to create forecasts, are of the type User Exit. They are based on the function module UPF_FORECAST_EXIT_INIT and UPF_FORECAST_EXIT_EXECUTE.

Sales: Exponential Smoothing with Trend and Season (4TPM8117)

This planning function creates forecasts of the type Exponential Smoothing, taking trend and season into account. It uses the forecast model 24 (Trend/Seasonal Exponential Smoothing (Mult. Season Comp.)) and creates forecasts for the key figure Billing Quantity in Sales Units (0INV_QTY). Values for this key figure with the plan version Current Forecast (CAF) are used as a reference. The reference period is defined by the variable FISCP02 (see also Variable for Definition of Periods). The planning function automatically corrects outliers, optimizes smoothing parameters, and automatically factors out missing values from the model calculation.

The parameters are set as follows:

Parameters

Description	Parameters	Data element	Selection
Version	0VERSION	UPX_Y_VERSION	CAF
Forecast strategy	11FCSTR	UPX_Y_VERSION	24
Periods/season	12PERIO	PERIO	12
Outlier correction	13OUTL	UPF_Y_OUTLIER_CORRECTION	X
Factor optimization	21OPT	UPF_Y_FC_OPT	X
Ignore null values	DISREG	UPF_Y_DISREG	X
Key figure (forecast)	KYFNM01	UPC_Y_KYFNM	0INV_QTY
Key figure (reference)	REFRATIO	UPC_Y_KYFNM	0INV_QTY
Reference period variable	TMVAR	UPC_Y_VARIABLE	FISCP02

 

Sales: Exponential Smoothing of Automatic Model Selection (4TPM8137)

This planning function offers Automatic Model Selection: the system compares constant, trend, seasonal and trend/seasonal models and selects the most suitable model for the forecast. The parameters generally have the same structure as the planning function described above: the planning functions differ only in terms of the forecast model used.

Automatic model selection is particularly useful when you enter the forecast. Bear in mind that the parameter definition affects performance.

Sales: Linear Regression (4TPM8167)

This planning function serves as an alternative to the forecast model Trend/Seasonal Exponential Smoothing (24) and allows you to use the forecast model Linear Regression (31). The parameters generally have the same structure as the planning function described above: the planning functions differ only in terms of the forecast model used.

The forecast models mentioned each require a particular number of historic values:

§         Constant model: 1

§         Trend model: 3

§         Seasonal model: 1 season length (defined in parameters)

§         Trend/seasonal model: 2 season lengths (defined in parameters)

Bear in mind that the results quality of complex forecast models and of the trend/seasonal model is largely dependent on the length of the reference period. If, for example, a trend/seasonal model has only two reference season lengths, the seasonal component can only be determined using two reference values. 

Net Revenue: Calculation by Formula (4TPM8127)

The planning function type Formula enables you to calculate the net revenue using the billing quantity that has been forecast. As well as the billing quantity, the planning function uses the average price (actual net revenue / actual billing quantity) in the calculation.  The three months prior to the current fiscal year period (FISCP01) are used in the calculation of the average price. If there are no values for this period, the previous three months are used. If there are no values for this period either, the planning function accepts an average price of zero and calculates a net revenue of zero.

The changeable fields in this planning function are Fiscal Year/Period, Key Figure Name and Version.

Log display (4TPM8147)

After you have created a forecast using one of the aforementioned planning functions, this planning function gives you an overview of the following points:

·        Statistical quality criteria

·        Number of determined outliers

·        Selected smoothing parameters

·        Selected forecast model (with an automatic model selection)

The planning function type User Exit is based on the function modules  UPF_FORECAST_LOG_DISPLAY and UPF_FORECAST_LOG_DISPLAY_INIT.

Copy CMP Version to CAF (4TPM8197)

The planning function type Copy allows you to copy plan data from the plan version Business Plan (CMP) to the plan version Current Forecast (CAF). The variable FISCP03 contained in the planning level ensures that only future business periods are copied.

Existing values can be overwritten during the copying process, so we recommend that you execute this planning function once a month after you have created a forecast.

Planning Sequences

The planning level Forecast (4TPM8107) provides the following planning sequences:

·        Exponential Smoothing: Automatically with Log (4TPM8157)

·        Exponential Smoothing: Trend/Season with Log (4TPM8158)

·        Linear Regression (Trend) with Log (4TPM8159)

The planning sequences mentioned each combine a forecast model with the log display. They are used in the planning folder Forecast (4TPM8100) to display the relevant log automatically after creation of a forecast (see also Planning Folders).