The AI-driven Smart Visualization engine draws the visual using the most appropriate visualization type, given your selected data. This includes deciding on the chart type (including grids) and position of the data elements. The engine is used by Smart Discover and Natural Language Querying to select visualization types automatically as you build your visual.
Note: The Discover Lite and Discover Pro apps also use the Smart Visualization engine to offer a visualization type "override" based on your current data. In these apps, the engine is not enabled while you build your discovery. You can, however, select a visualization type from an Auto Recommended list that is made available on the right-click.
AI Mechanics
To better understand its mechanics, the following elements are taken into account:
- The number and type of metrics (values) selected.
- The number and type of hierarchies (dimensions) selected.
- The size of the number of elements (members) per hierarchy. This helps plotting very large dimensions vs very small dimensions and so on.
- The types of hierarchies (regular, time based, geospatial).
- Any filters and sorts applied to hierarchies.
- Any slicers (data filters) selected.
With these elements, the engine will plot the data in the most appropriate fashion. This includes:
- Decisions related to data positions:
- Categories (X-axis) and series (legend).
- Rows and columns.
- Trellising (multi-charts).
- Decisions to Visuals type:
- Grids.
- Cartesian charts (column, bar, line, and so on).
- Segment charts (pies, donuts, and so on).
- Plots (scatter, bubble, and so on).
- Advanced charts (circle packing, tree maps, and so on).
Overriding the Automation
Depending on the tool using the Smart Visualization engine, these choices can be overridden by a different chart type. However, the engine will automate the LAYOUT of the data elements for the given visualization. Click below to learn about overriding the Auto Recommended visualization type in:
To override the LAYOUT (and for more granular control of the selections), you should open the discoveries using Discover Lite or Discover Pro if you have access to these tools.