How to display non-aggregated metrics in charts

Dataiku Administrator, Dataiker, Alpha Tester Posts: 88 Administrator

Some types of data visualizations plot raw (or non-aggregated) values, while other visualizations calculate some kind of aggregation. A scatterplot, for example, plots raw values, while a histogram plots bin counts.

The type of chart determines whether displaying raw values is possible.

Native Charts

The drag-and-drop interface for creating charts in Dataiku DSS has five native chart types:

  • Basics
  • Pivot tables
  • Scatters
  • Maps (geographic)
  • Others (such as boxplot or 2D distribution)

DSS natively supports using non-aggregated values in scatter plots. For creating other types of charts, it is possible to plot non-aggregated values when grouping by a dimension with unique values. This is often most useful when the data is already in a summarized format.

In cases where the grouping variable does not have unique values, be sure to select the correct aggregation through the interface. For example, should DSS sum the values? Take an average? Or plot the minimum or maximum value in the group?


Advanced Visualizations

More advanced visualizations can be created by developing web apps within DSS. Several types of web apps are possible:

  • HTML + CSS + JS + optional Python
  • Shiny (R)
  • Bokeh (Python)

You can also publish visualizations from various Python and R libraries (such as matplotlib or ggplot2) as static insights on dashboards. For more information on embedding charts into dashboards as static insights, please see the relevant documentation for Python or R libraries.


What’s next?

Try out some hands-on tutorials for creating charts in Dataiku DSS.

Setup Info
      Help me…