Smart Insights: Top Contributors to Your Data Point

Smart Insights can show you the main contributors to your data point.

How Does the Algorithm Work?

What's happening in the background is that Smart Insights performs a live analysis of your data point to highlight its most interesting dimensions. It also obeys any applied filters, and user authorizations. The Smart Insights algorithm ranks the highlighted dimensions in order, depending on how much their member values deviate from the average with regard to the selected dimension. You can see the top 5 dimensions and their top 10 member values displayed as top contributors.

Customizing Your Results

As a story creator, you can customize the top contributor Smart Insights results, to make sure they are as relevant as possible to your business context. If certain dimensions or dimension members aren't interesting to you, you can exclude them. Top contributors found for other dimensions are then shown in the Smart Insights panel instead. The selected dimensions or dimension members aren't factored into your top contributor results. The exclusion of these dimensions or dimension members persists for the lifetime of all charts connected to the same model in your story, until you select to include them again.

The excluded dimension members are added to the Others category in the summary chart in the Smart Insights panel, even if there are less than top 10 contributing dimension members.

How to Customize Your Top Contributor Results

  1. Select Edit mode within your story. If the toolbar is collapsed, choose Start of the navigation pathFile Next navigation step  (Edit Story) Next navigation step Smart Insights SettingsEnd of the navigation path. If the toolbar is expanded, select Start of the navigation pathFile Next navigation step  (Edit Story) Next navigation step Smart Insights SettingsEnd of the navigation path.

    The Smart Insights Settings dialog opens.

  2. Identify the relevant model from the list, and select + Select Dimension Members to open the list of dimensions you could potentially exclude from your top contributor Smart Insights results.

  3. Select the relevant dimension from the list.

    The Select Members to Exclude from Smart Insights for <selected dimension> dialog appears.

  4. Select All Members to exclude the entire dimension.

    Your selection All Members then appears in the dialog box with a strikethrough, to indicate that the entire dimension is excluded from your top contributor Smart Insights results.

  5. To exclude specific dimension members, select instead the relevant checkbox beside the dimension member name. You can also use Search to find specific dimension members.

    The list of selected dimension members then appears in the dialog box with a strikethrough, to indicate that those specific dimension members from that dimension are excluded from your top contributor Smart Insights results.

  6. Select OK.

    Now, when you run Smart Insights, the dimensions or dimension members you selected to exclude, won't be factored into your top contributor results. The top five contributors found for other dimensions are shown in the Smart Insights panel instead.

  7. To include dimensions or dimension members again, so that they're factored into in the top contributor Smart Insights results, choose Start of the navigation pathFile Next navigation step  (Edit Story) Next navigation step Smart Insights SettingsEnd of the navigation path, and deselect the dimensions or dimension members.

  8. Select OK.

    When you run Smart Insights again, these dimensions or dimension members are again factored into your top five contributor results.

Hierarchical Data

All dimension members, not just at the leaf node level, are considered by Smart Insights as potential top contributors to your selected data point. The dimension can even have multiple hierarchical levels, or already appear in your chart and still be considered. If a level in your dimension hierarchy has only one member, Smart Insights discards it as uninteresting. If Smart Insights calculates the same score for two levels in your hierarchy, it favors the level closer to the root of your dimension hierarchy. Every member in a hierarchy is looked at.

To see at what level in your hierarchy a dimension occurs, simply hover over the dimension in the chart or the title of the Smart Insights panel.
Note
  • If a dimension in your hierarchy has over 200 K nodes, the top contributor insight for that specific dimension is based on the level of the leaf node, not the other hierarchical levels.

  • If your dimension hierarchy levels run deeper than 5 five levels down from the root, the graphical display of the hierarchy is truncated to only show the bottom four levels.

  • If you apply a filter to a specific dimension in a dimension hierarchy, the top contributor insight for that specific dimension is based on the level of the leaf node, not the other hierarchical levels.

  • Smart Insights won't return top contributor insights if the returned result sets for the top contributor queries from the backend is more than 1 million rows.

Smart Insights Panel

The Smart Insights panel highlights the contributing factors (dimensions and dimension members from your model), to the data point you selected. Each contributing factor is explained with a textual insight and a chart.

If your selected data point has a large number of generic dimensions, you can see two charts within the top contributor section of the Smart Insights panel.

The first chart shows you the top 10 (or less) contributors to your selected data point, sorted from largest to smallest. If your selected data point has less than 10 dimension members, this chart also displays an average line. You can drill into the insights in this chart to generate another set of top 10 contributors to your selected data point. Unassigned members aren't considered as top contributors to your selected data point.

When there are greater than 10 contributors, or if there are any unassigned values, the second chart shows a summary of how the total measure value of your data point breaks down into different categories. Depending on your data model, the summary chart displays a combination of these categories:
  • Top 10 contributors (assigned values)

  • Others (other assigned value contributors beyond the top 10)

  • Unassigned values (if there are any to show)

Note
  • When working with models on a live SAP HANA connection, unassigned values are considered as top contributors to your selected data point, and so don't appear in the summary chart.

  • The summary chart can't currently be copied to your story.

  • Working with variance based on a calculated measure is not supported for top contributor insights.

Drilling Down into Your Data

Drill into the insights in your chart and see another chart offering further insights. Want to know more? Simply select another data point of interest within the chart, and select the (Smart Insights) again. The Smart Insights panel offers more insights. Go deeper into your data, and watch as other insights in the panel adapt to your new data exploration path.

Want to go back to the previous panel view before you started to explore your insights? Select (Back) in the Smart Insights panel.