Add a Language Narrative Visualization

The Language Narrative visualization provides insights into your company's data. These insights can be in the form of breakdown or trend.

LiveLabs Sprint

Topics:

• About the Language Narrative Visualization

• Create a Language Narrative Visualization

About the Language Narrative Visualization

The Language Narrative visualization creates natural language descriptions of the attributes and measures in your dataset. The descriptions provide insights into your company's data in the form of breakdown or trend.

For example, if you create a workbook with Sample Sales and you drag the T00 Calendar Date time dimension and the 1-Revenue measure to the canvas and then select the Language Narrative visualization, you'll see the following description:

Description of the illustration lang_narrative_viz_rev_cal.jpg

The natural language description capabilities are provided by the integration of Oracle Analytics Cloud with its artificial intelligence technology partner Yseop.

Supported Data Element Combinations

You can use the Language Narrative visualization with the following combinations:

• Two attributes and one measure.
• One attribute and two measures.

The attributes that you select can be regular columns or time dimensions. And columns may contain long character names and special characters.

Use these tables to understand the type of analysis that Oracle Analytics Cloud performs based on the attribute (regular or time dimensions) and measure combination that you selected.

Two Attributes and One Measure Combinations

Item	Dimension 1	Dimension 2	Measure	Chronological Order	Unrelated	Result
One regular or time dimension One measure	Regular or time	-	Regular	Yes	Yes	Trend analysis for the time dimension. Breakdown analysis for regular column. When you initially create the visualization, these analyses are the default. If needed, you can choose the other analysis type (trend or breakdown).
One regular or time dimension One measure	Regular or time	-	Regular	No	Yes	Trend analysis for the time dimension. Breakdown analysis for the regular column. When you initially create the visualization, these analyses are the default. If needed, you can choose the other analysis type (trend or breakdown).
Two regular or time dimensions One measure	Regular or time	Regular or time	Regular	Yes	Yes	Trend analysis for the first dimension, if it's ordered (namely, is a time dimension), and breakdown for the second dimension. Breakdown analysis if both dimensions are regular columns. Time dimension is trend analysis and regular column is breakdown analysis.
One time dimension One regular dimension One measure	Time	Regular	Regular	Yes	Yes	Trend analysis for the first dimension. Breakdown analysis for the second dimension.
One regular dimension One time dimension One measure	Regular	Time	Regular	Yes	Yes	Breakdown analysis for only the first regular dimension.

One Attribute and Two Measures Combinations

Item	Dimension	Measure 1	Measure 2	Chronological Order	Unrelated	Result
One regular dimension Two measures	Regular	Regular	Regular	No	Yes	Breakdown analysis for the first and second measures individually, based on dimension.
One time dimension Two measures	Time	Regular	Regular	Yes	Value verses reference	Trend analysis for the first measure with a value comparison to the second measure.

Other Limitations

Note the following limitations for Language Narrative visualizations:

• The text in the visualization can be displayed in English or French. Other languages aren't supported. Oracle Analytics Cloud maps French locales (fr and fr-CA) to the French language, and maps all other locales to English.
• Presentation mode and embedding items such as dashboards aren't supported.
• The color change and data brushing functionality isn't applicable to this visualization type.
• Data actions aren't applicable to this visualization type.
• The Use as Filter option isn't available for this visualization type.
• Oracle Analytics Cloud can't generate a natural language analysis if your visualization contains two attributes and the second attribute has more than 50 elements.

Create a Language Narrative Visualization

Use the Language Narrative visualization type for descriptions of the attributes and measures in your dataset. The descriptions provide insights into your company's data in the form of breakdown or trend.

If you haven't already created a workbook, you need to create one. See Begin to Build a Workbook and Create Visualizations.

To create a Language Narrative visualization, must select either two attributes and one measure, or one attribute and two measures. See About the Language Narrative Visualization.

1. On the Home page, hover over a workbook, click Actions , then select Open.
2. Click Visualize and double-click the Language Narrative visualization type to select it.
3. Display the Data pane and drag data elements to the canvas. You must select either two attributes and one measure, or one attribute and two measures.
4. Make adjustments to see the data from different angles:
   • Change the data elements in the drop targets. Remember, you can only use the supported measure and attribute combinations as described in Step 3.
   • If the visualization has a calendar date or time stamp column on the outer edge, then in the Properties Pane's General tab, change the analysis type. Select Trend for a trend analysis. Trend analyses are ordered and can't be disjointed. Select Breakdown for a breakdown of the data based on the attributes and measure you chose.
   • In the Properties Pane's General tab, use the Level of Detail field to change the level of detail displayed. By default, you'll see your data at 7, the most detailed level, but you can move the slider down to display the data in larger chunks.
   • In the Properties Pane's Values tab, use the Meaning of Up field to indicate if the attribute or measure value's definition is good or bad. For example, if you select the Revenue measure, then set the Meaning of Up field to Good. If you select the Average Order Process Time measure, then set the Meaning of Up field to Bad.