Link Dimensions Across Multiple Models

In SAP Analytics Cloud, you can create links between dimensions in multiple models.

You can link dimensions between models to create charts or tables that display data from multiple models.

Linked dimensions also let you create filters that simultaneously update all charts that include linked data. Filters on linked dimensions can be used at the story, page, and linked analysis level.

Note
  • The attributes of a dimension must match the attributes of the dimension it is linked to:

    • Description to Description
    • ID to ID

    If the attributes don't match, information might not be returned when a filter is created, and a “(No Value)” flag will appear for an unmatched dimension member.

  • Chart filters don't affect linked dimensions, because they're applied to a single chart.
  • Some dimension links may not apply filters to all charts. This happens when filters applied to one model can't be translated to meaningful filters on a second model.
  • For BW remote models, consider that the Description might be language-dependent.
  • Sorting doesn't work with linked dimensions.
Restriction
The Show Totals feature is not supported with linked dimensions.

How to Create Linked Dimensions

  1. Use one of the following methods to open the Linked Dimensions dialog:
    • Select (Link Dimensions) from the menu bar.

      Note
      If there are already linked dimensions in the story, they will be listed. To use the existing link, select (Edit Link). To create a new link, select Add Model Link.
      1. Select a chart or table.
      2. In the Builder tab, in the Data Source area, select Add Linked ModelsCreate Model Link.
  2. Select a model from the list of models included in your story, or in the drop-list select Add Model to search through the models you have access to.

  3. (Optional) To change how the dimensions are displayed in the Select a model area, select Start of the navigation path Next navigation step Data SamplesEnd of the navigation path and then select an attribute:

    • Description
    • ID
    • ID and Description
    Note

    By default, only descriptions are retrieved. For dimensions that don't have a Description attribute, No Values will be displayed. Selecting ID or ID and Description gives a better indication of the true values available.

    Note

    For dimensions that have blank or empty strings for Description (""), the value of the ID attribute will be displayed. This can sometimes lead to an incorrect perception that two dimensions can be matched on their Description attribute when in fact the ID attribute should be used.

  4. Choose one or more dimensions.

    Tip
    When you hover over a dimension, you can preview the dimension values.
  5. Choose a second model and select one or more dimensions.

    Note
    The attribute of a dimension must match the attribute of the dimension it is linked to. If it doesn't match, when a filter is created, information may not be returned.
  6. (Optional) In the Matched Dimensions area, select a dimension, select (Link Attribute) and then choose from the linking options.

    Option Filtering Across Models Blending Data

    Link On: ID or Description.

    This feature is available on non-hierarchical dimensions and imported data source dimensions with level-based hierarchies.

    Selecting ID or Description is not possible for dimensions with parent-child hierarchies.

    For filtering across models, the ID attribute is used to propagate filters from source models to target models.

    For blending, the Description attribute is used to match members between the linked dimensions.

    Link On specific level-based hierarchy

    This option is available on dimensions with level-based hierarchies.

    • For imported data sources, link on the dimension ID and the individual hierarchy properties, rather than the hierarchy itself.

    • The dimension can only be linked to another dimension with a level-based hierarchy that has the same number of levels. For best results, the levels should contain similar or matching information.

      For example, it is possible to link a two-level hierarchy representing Country-City to another two-level hierarchy representing Country-City or to a two-level hierarchy representing Continent-Country. However, only the first link would yield expected results.

    Not applicable

    Link on matching hierarchies

    This option is not available for Live Universe models.

    This setting is ON by default for any links created starting with version 2020.15 and OFF by default for any content created prior to 2020.15.

    Matching is done based on the hierarchy name/id and assumes that hierarchies with the same name have matching structure/members. For cases where this assumption is true, it is recommended to turn ON the option.

    When the setting is OFF, filtering across models propagates filters on the default hierarchy: this can cause poor performance for source filters that use the non-default hierarchy.

    When the setting is ON, filtering across models propagates filters on the matching hierarchy (provided one exists) no matter which hierarchy is used in the source filter.

    When the setting is OFF, blending uses the default hierarchy in the secondary sub-query: this may not match members of the parent-child hierarchy set on the primary sub-query.

    When the setting is ON, blending uses the matching hierarchy (provided one exists) in the secondary sub-query based on the parent-child hierarchy set on the primary sub-query.

    Exclude descendants

    This option is available on dimensions with parent-child hierarchies and Live HANA model dimensions with level-based hierarchies.

    If the hierarchies have identical structure/members, ensure that the filtering is done using the currently selected members only. Be sure to select the Exclude descendants option.

    If the hierarchies aren't identical, then you might need to filter on descendants to get the right filter result, at the cost of slower performance.

    Not Applicable

    By default, the Description attribute is used to match members between the linked dimensions. However, for non-hierarchical dimensions it is possible to change the attribute that is used for linking.

    Note
    Filtering across models on hierarchies can be performance intensive. There are a few things to consider:
    • If the hierarchies aren't identical, then you might need to filter on descendants to get the right filter result, at the cost of slower performance.

    • If the hierarchy is a level-based hierarchy for an import data model, we suggest that you link on ID and the individual hierarchy properties, rather than the hierarchy itself.

  7. Select Set.

  8. Review the links and if necessary, edit them or select Add Model Link to add more links.

  9. When finished, select Done.

How to Edit Linked Dimensions

  1. Use one of the following methods to open the Linked Dimensions dialog:
    • Select (Link Dimensions) from the menu bar.

      Note
      If there are already linked dimensions in the story, they will be listed. To use the existing link, select (Edit Link). To create a new link, select Add Model Link.
      1. Select a chart or table.
      2. In the Builder tab, in the Data Source area, select Add Linked ModelsCreate Model Link.
  2. Choose an existing model link and select (Edit Link).

    If you want to add a new link instead, then select Add Model Link.

  3. Change the models or dimensions used, and then select Set.

    Select Done.

All charts or tables that include the linked dimension will update simultaneously. In the Builder tab, linked dimensions are indicated by the (Link) icon.

Difference Between Filtering Across Models and Blending

When filtering across models, you link dimensions together.

When blending data, you combine data from more than one model into a chart or table as a part of a story.

In both cases, however, the first step is to link the common dimensions of one model with the other.

For more information about filtering across models, see Filter Across Models.

For more information about data blending, see the following topics: