Demand Planning cluster-level configuration¶

Cluster-level configuration defines how forecast is calculated for one material-cluster and location-cluster combination within an execution profile.

It is different from cluster registration, which defines which materials and locations belong to each group.

Demand Planning cluster-level configuration

Dark-theme demo: select the profile and cluster pair before reviewing forecast parameters or generating a preview.

Path in the platform¶

Demand Planning > Configuration > Cluster-Level Configuration

Grain and key¶

One configuration is identified by:

demandPlanExecutionProfileId × materialClusterId × locationClusterId

Changing the parameters changes future simulations and runs for that combination. It does not rewrite Demand Plans already generated.

Required fields¶

Field	Meaning
`demandPlanExecutionProfileId`	Execution profile owning the configuration
`materialClusterId`	Material cluster receiving these parameters
`locationClusterId`	Location cluster receiving these parameters
`demandPlanningGeneralParameters`	General-parameter block
`demandPlanningForecastParameters`	Forecast-parameter block
`statisticalModel`	Statistical model selected explicitly
`splitModel`	Split model selected explicitly

Optional general parameters¶

Field	Meaning	Default Value When Not Provided
`executeDemandPlan`	Whether this cluster pair participates in Demand Planning	`true`
`uomId`	Unit used to calculate and present the forecast	Demand Planning unit resolved from profile/global parameters
`roundToSalesUnit`	Whether forecast is rounded to the material's sales unit	`false`
`considerHistoricalSalesOfInactiveDfus`	Whether inactive DFUs remain eligible for model history	`true`
`generateForecastForDiscontinuedMaterials`	Whether discontinued materials may still receive forecast	Effective value in Global Planning Parameters
`materialAggregationType`	How the material dimension is calculated	`Top-Down`
`locationAggregationType`	How the location dimension is calculated	`Top-Down`
`daysSalesHistory`	Historical window available to the statistical model	Statistical-history window in Global Planning Parameters

Optional forecast parameters¶

Field	Meaning	Default Value When Not Provided
`daysMovingAverageModel`	Historical window used by Moving Average models	`120` days
`daysTopDownSplit`	Historical window used by the `Historical Sales` split	`120` days
`alpha`	Level smoothing for Exponential Smoothing or Holt-Winters	Automatic selection
`beta`	Trend smoothing for Holt-Winters	Automatic selection
`gamma`	Seasonal smoothing for Holt-Winters	Automatic selection

Availability by edition¶

Capability	Community	Enterprise
Configuration by profile and material/location cluster pair	Available	Available
Top-down or bottom-up by dimension	Available	Available
Community statistical-model catalog	Available	Available
Split by historical sales	Available	Available
Use the profile's approved AutoFit model	Unavailable	Available
Additional statistical and foundation models	Unavailable	Available
Forecast-proportion or hierarchical split	Unavailable	Available
Stockout treatment, history cleansing, and event uplift	Unavailable	Available
Budget as Forecast and internal regressors	Unavailable	Available

See the exact model availability in Statistical Model and Split Model.

How to configure¶

Select the execution profile.
Select one location cluster and one material cluster.
Enable or disable execution for that pair.
Choose top-down or bottom-up independently for material and location.
Select the statistical model and review its visible parameters.
If either dimension is top-down, select the split model and its reference window.
Confirm the unit, history, inactive/discontinued treatment, and rounding.
Save the configuration.
Generate a forecast preview and inspect series, bias, error, and split before the official run.

What is configured elsewhere¶

Decision	Correct configuration
Which records belong to each cluster	Forecast clusters
Source, bucket, plan horizon, and edit window	Execution profile
Generation and comparison of AutoFit candidates	Cluster-level AutoFit
Collaboration stages and editable key figures	Demand Planning workflow

Dependencies¶

a valid execution profile;
material and location clusters with complete allocation;
sales history and an eligible historical window;
units of measure;
a split rule whenever at least one dimension is top-down;
an approved Enterprise AutoFit model before enabling profile AutoFit.

Used by¶

forecast preview and parameter simulation;
official Demand Planning execution;
statistical baseline generation and split to DFUs;
Enterprise AutoFit winner resolution;
Demand Plan analysis and collaboration.

Validations¶

profile, material cluster, and location cluster are present;
general and forecast parameter blocks are present;
statistical model and split model are selected explicitly;
historical windows are positive when filled;
selected model and split are available in the current edition;
Historical Sales split has a positive reference window;
an Enterprise model is compatible with its split and advanced parameters;
unit and all referenced objects exist.