Skip to main content

Aggregation and entity types

Arria's NLG Apps algorithms analyze measures (e.g. revenue) in the context of dimensions (e.g. city and month).

The first iteration of NLG Apps analyzed measures aggregated by Sum: our algorithms were designed to calculate totals, variance, standard deviations, medians, etc. from summed measures and report on them by dimension instances. For example:

Total Revenue for New York in February was $30M, up from $25M in January.

The analysis and calculations performed by our algorithms were not suitable for precalculated values like percentages and ratings, nor for measures aggregated by anything other than Sum. For example, reporting on the sum total of customer credit ratings is not meaningful; rather, we should report only on averages, medians, etc. like this:

For all Customers, the average Credit Rating is 2.78, and the median Credit Rating is 2.9.

We have since adapted our NLG Apps algorithms to adjust the analysis and calculations performed according to a measure's selected characteristics. These characteristics include aggregation type (set in Qlik Sense) and entity type (set in Arria's extension).

Arria for Qlik Sense 3.1 added support for the following aggregation types:

  • Avg, Min, Max, and Count

In addition, the following entity types have been added:

  • Percentage, Rating, and Ratio

 

You can find details on all of our apps in the NLG Apps Directory, including which combinations of aggregation and entity types they each support.

For guidance on how to configure and generate NLG Apps narratives, see our tutorial and reference documentation.

 

Note

Qlik Sense does not provide a way to add non-aggregated measures to visualizations. Therefore, at present, raw data that comprises percentages, ratings, and ratios cannot be analyzed by Arria for Qlik Sense.