Oracle Analytics Cloud and Server Idea Lab

Welcome to the Oracle Analytics Community: Please complete your User Profile and upload your Profile Picture

Histogram Viz Type

Needs Votes
236
Views
0
Comments

Organization Name (Required - If you are an Oracle Partner, please provide the organization you are logging the idea on behalf of):

Worthington Industries

Description (Required):

Currently the only way to create a histogram is by using a bar chart and assigning bins to the dataset based on some criteria. The options for binning are either to use the built-in BIN function, which has some short-comings, or to do a custom expression that assigns a bin value (For example, <FLOOR([MEASURE] / 5) * 5>, effectively creates a bin width of 5); however, while this custom expression allows for better binning than the BIN function, it still has some of the same shortcomings:

  1. Requires custom calculations for both the x and y axes (Bin assignment and count, respectively).
  2. X axis is limited to available data, and is not continuous; in other words, bins with no count are excluded.
  3. No ability to set custom reference lines, nor reference lines against the bin axis (X axis).
  4. No simple way to plot a distribution curve.

Use Case and Business Need (Required):

Due to the aforementioned gaps between the current bar charts and a true histogram distribution chart, I would like to see histogram added as its own chart type. Some desired functionality/properties of the chart would be:

  1. Drag and drop a measure and an attribute onto the canvas, picking Histogram as the viz type, and then using viz properties to allocate a desired bin size (Would need this bin size to be compatible with the upcoming presentation variables/parameters, and not just a constant value); likewise, the measure should automatically aggregate to a count of records, regardless of semantic model or custom dataset defaults.
  2. There should be an option in the viz properties to choose whether null bins are displayed or not; for many statistical applications, most specifically process capabilities and ANOVA, a full view of the data distribution, including nulls, is required.
  3. Reference lines should be fully customizable, bases on expressions, and against either axis (Separate idea entered for this, as it should apply to many/all chart types, not just histograms: https://community.oracle.com/customerconnect/discussion/665405/fully-customizable-reference-line-locations/p1?new=1)
  4. Distribution curve should be an available option in the Analytics tab of the viz properties.

Enhancement Request / Service Request:

9
9 votes

Needs Votes · Last Updated