Configure Decomposition Groups

A decomposition group is a container for the measures that you use as causal factors in a forecasting profile that's based on Bayesian machine learning.

Causal factors enable forecasting methods to understand the variation in historical demand and produce an accurate and adoptive forecast. Decomposition groups let you organize measures that have similar effects on a forecast.

Several predefined causal factors (predefined measures) are available for meeting commonly found forecasting needs. You can also use user-defined measures as causal factors. You can easily view and modify the information in causal factors using tables. Causal factors are the best way for you to add information that drives demand into the forecasting process and thereby improve it.

For your user-defined forecasting profile, you can add, edit, or delete decomposition groups. You can also activate and deactivate all causal factors in a decomposition group using its check box.

In a demand or demand and supply plan, you can decompose the forecast by the causal factor group. When you select the Include details of causal factors check box on the Parameters tab in the Run Plan dialog box, the duration of the plan run increases significantly. For this reason, decompose the forecast by causal factors preferably when you simulate the demand from a table for a subset of the plan that's based on the table selections. By doing so, you can perform a focused analysis of the causal factors. Include the details of the causal factors only when you want to analyze the results and not every time you run the plan.

Note: Causal decomposition isn't supported for the Causal Winters (B), Croston for Intermittent (F), Multiplicative Monte Carlo Intermittent (K), Regression for Intermittent (J), Naive (N), Moving Average Naive (O), and Holt Naive (T) forecasting methods. No decomposed values are generated for any items that are forecast with these forecasting methods.

For information about causal factors, refer to the white paper titled "Demand Management Forecasting Causal Factors" that's available in Document ID 2551482.1 on My Oracle Support.

Create a Decomposition Group

Perform these steps to create a decomposition group:

1. On the Decomposition Groups tab in your forecasting profile, click Actions > Create.

2. In the Create Decomposition Group dialog box, do the following:

   • Provide a name and description for the decomposition group.

   • From the Available Measures pane, move the required measures to the Selected Measures pane.

     The available measures are those that are enabled for the work area in which you create the decomposition group. Ensure that the selected measures are in the measure catalog of the plan that uses the forecasting profile.

3. Click OK.

Configure Causal Factors in a Decomposition Group

Perform these steps to configure the causal factors in a decomposition group:

1. On the Decomposition Groups tab in your forecasting profile, expand the decomposition group.

   The expanded list details the available causal factors.

2. Select the check boxes for each causal factor:

   • Short: Use this setting to assign the causal factor to the Causal Winters (B), Logistic (G), Regression (R), Regression for Intermittent (J), and Transformation Regression (L) forecasting methods that use all assigned causal factors. These forecasting methods use a limited set of causal factors and aren't efficient at determining which causal factors are useful. You select this check box for most causal factors.

   • Long: Use this setting to assign the causal factor to forecasting methods that use an extended set of causal factors and are efficient at determining which causal factors are useful. These forecasting methods are Auto Regressive Logistic (A), Combined Transformation (E), Modified Ridge Regression (M), Multiplicative Monte Carlo Intermittent (K), and Multiplicative Monte Carlo Regression (C).

   • Multiplicative: Use this setting to assign the causal factor to the Dual Group Multiplicative (D) forecasting method. If you have enabled this forecasting method, each causal factor of your forecasting profile can use the Group One or Group Two value of the list. Each of these values should be selected for at least one causal factor.

   • Not Seasonal: Use this setting to assign the causal factor to the Auto Regressive External Inputs (X) and Auto Regressive Integrated External (V) forecasting methods. The only causal factors that should be assigned this setting are those that aren't predictable functions of time. For example, price varies randomly with time and can be assigned this setting.

   • Fill Missing: Use this setting for causal factors when there are missing values for dates, and you want to have values for these dates. The values that are entered for a causal factor are based on values of the same causal factor for other dates. Use this setting for causal factors, such as those for price, that should always have a value.