Tuesday, April 18, 2017

Data Visualization - Smart Insights


Smart Insights have been available in Oracle Data Visualization Desktop for a few releases and now available in Oracle Analytics Cloud (OAC)

So what are Smart Insights?
• They provide an at-a-glance assessment of your data
• Allows analysts to quickly understand the information the data contains
• You can easily see how measures are distributed in various attributes/dimensions
• Provides starting point for further data analysis

In looking at tabular data it is difficult to see patterns and distribution of measures across dimensions



When we look at the same data via Smart Insights we can see how the data is distributed across attributes/dimensions.
So how does one access Smart Insights?

First add a data source to a project.  Then switch to Prepare Mode. Finally switch the view from Data to Visual.

The Visual Mode creates a series of simple data visualizations.  The initial views are the number of rows of data distributed across all the attributes/dimensions in the data set.


The view can easily be changed from row count to other measures/facts in the data set by changing the Summarize by.
Additional properties that help with analyzing the data is flagging whether show null rows in an attribute/dimension.  The include others option is useful when the data has many different values in an attribute since the number of values on an axis also referred to as binning is limited.


The Binning of values on the X and Y Axis follow these
• Number of Bars Depends on Data Distribution
• Normally 10 bars are shown and all other data is displayed in a bar called Other
• If 20% or more of the data falls into the Other the system will break that data into the number of bars needed



The style of the visualizations align to the type of data in the attribute/dimension.
• Non-numeric or Text - Horizontal bar chart
• Date and Time - Line chart
• Numeric - Vertical bar chart



From the samples shown throughout this post one can see how beneficial they are to easily understanding the data, finding initial patterns and providing a starting point for further data analysis.

3 comments: