Defining the Model Scope

Define the slice definition, or model scope, for both historical data and future data for Advanced Predictions.

Define the model scope for the Advanced Prediction:

  • On the Historical Data page, you select members from each dimension to define the location of the historical data that should be used to train the ML model for the advanced predictions. See Defining the Model Scope for Historical Data.
  • On the Future Data page, you define where you want to store the output from predictions and the data slice for drivers in future periods. See Defining the Model Scope for Future Data.

Defining the Model Scope for Historical Data

To define the model scope for Historical Data:

  1. In the Define Model Scope section, first determine which dimension has the measure you want to predict. Click the down arrow Down arrow icon next to this dimension name.

    The dimension is now listed in Select Output to Predict and Select Drivers as Input .

  2. In the Define Model Scope section, for the remaining dimensions, select the members that define the location of the historical data that should be used to train the advanced prediction model.

    To select the members, click a dimension, and then use the member selector to select the members. You can also use functions to select members.


    Define model scope example

    Member selector to define the model scope

Defining the Model Scope for Future Data

To define the model scope for Future Data:

  1. In the Define Model Scope section, select members from each dimension to define where to store the prediction data.

    Note:

    Member selections are pre-filled with the member selections you made for historical data, so you need to change only the specific members where future data exists and where the predictions are to be stored. For example, you might only need to change Scenario from Actual to Forecast.
  2. If needed, make any required changes to Select Output to Predict or Select Drivers as Input.
  3. Click Next.