Displaying highlights
The display of highlight facts uses advanced algorithms to analyze your data to improve understanding of it. Which element(s) contribute most to a value? Is there a correlation with another measure? ...
Artificial intelligence analyzes the dimensions and measures contained in your data model to provide you with relevant information related to the selected data to identify, for example, anomalies or trends.
💡Consult the Artificial Intelligence > Highlight page for configuring highlight facts.
Configuring the display of highlights
Selecting the dimensions and measures to analyze
By default, all dimensions and measures in the data model are selected for AI data analysis.
You can modify this selection to only include certain dimensions and measures. The selection will be applied by default to all flows (charts) using this data model.
To do this:
- In the Studio, edit the data model.
- In the Columns tab of the model's advanced configuration, select the dimensions and/or measures you wish to exclude from the highlight facts analysis.
- In the Properties section of the right panel, uncheck the AI Analysis box.
Activating the display of highlights
Highlights are disabled by default for all flows (charts).
To enable highlight facts for a chart:
- Select the chart in the Dashboard Editor.
- Go to the Properties section of the Settings panel on the right.
- In the Options tab, check the Show highlights option.
Displaying highlights in the dashboard
Simply click (see the Interaction and highlights section) on an element to display the highlight facts or key points.
Digdash then searches for relevant information related to the selected data among:
- measures present in the flow that have been selected for AI analysis (hidden or not)
- dimensions (and all associated hierarchy levels) that are not displayed on the flow and have been selected for AI analysis.
A window appears with the key points for the selected chart.
It specifies the selected exploration filters and lists the different highlight or key points identified, sorted by measure then by dimension.
In case of a large number of key points, a loading bar appears at the bottom of the window with a button to cancel the loading:
Interaction and highlights
By default, charts inherit the click interaction defined during their configuration in the Studio (if any). Consult the Configure interactions page for more details on chart interactions.
In this case, to display highlight facts, right-click on the data and then click Show highlights.
If you want to set highlight facts as the default click interaction, regardless of the case:
- Go to the User menu and click Default interaction.
- Below Default interaction choice, click Highlights.
You can also use the Ctrl + H shortcut. A message will then appear.
Types of highlights
Several types of key points can be identified.
Mega contributor 
The mega contributor refers to a dimension member that contributes at least 40% to a given measure (total sum).
The dimension must contain at least 5 members.
For example, in a store network, the United States accounts for 54% of revenue.
Top K contributors 
Top K contributors refer to the K best members of a dimension that contribute at least 40% to a given measure (total sum).
K ranges from 2 to 5. Its value depends on the number of members. By default, K equals 33% of the number of members.
Thus, to get a Top 2, a minimum of 6 members is required. For a Top 3, a minimum of 9 members; for a Top 4, a minimum of 12 members; and for a Top 5, a minimum of 15 members.
For example, in a store network, the Fruits, Gift Set, and Basket products account for 62% of revenue.
Pareto 
Pareto refers to the case where 20% of a dimension's members contribute to 80% of a given measure.
For example, in a store network, it is observed that the top 20% of best-selling products account for 80% of total revenue.
Correlation 
Correlation refers to the case where a measure (not displayed) evolves in the same way as the selected measure, according to a given dimension.
Correlation search is only performed when the selected measure has a Mega contributor, Top K contributors, or Pareto.
Correlation can be positive or negative and have different levels of significance: very strong, strong, or moderate.
The calculation of correlation thresholds is dynamic. The minimum threshold starts at 0.7 for 10 members.
As the number of members increases, the minimum threshold becomes more permissive, reaching 0.3 for 50 members or more.
For example, a very strong correlation is observed between the Revenue measure and the Quality Index measure.