5 Replies Latest reply on Apr 24, 2016 11:53 PM by Charles Zhang

    Data quality in data lookup measure

    MassimoSporchia

      Hi all,

       

      I have a measure with a Data Reference of type "Table Lookup" and when I query the result, if the value is present (query returning at least a row compliant with the "Behaviour" chosen) the data quality is Good, otherwise is Bad.

      Hence three questions:

      1 - Is there a way to manipulate data quality? For example, get the value from another column retrieved from the table?

      2 - I've never seen Suspect returned by a Table lookup: is it caused by the binary behaviour of a query? (it's either valid or not)

      3 - I'm using AF Client 2014 and I'm able to set the returned value to 0 if the query returns no row or null, shouldn't the value's quality be Bad/Suspect instead of the Good I'm actually receiving?

       

      Thanks in advance!

        • Re: Data quality in data lookup measure
          Rhys Kirk

          Depends how you structure your AF hierarchy.

          If you want an alternative value then you could return both columns in AF Attributes and have a parent AF Attribute that decides which value to use. Or you could embed that logic into a SQL view or equivalent.

          The "set returned value to 0" is a substitution of value and the "0" value is considered good - at least for my interpretation. How you handle the quality further downstream really is up to you.

          For example, you could return all values (raw) and then have an AF Analyses that outputs a Cleansed Value AF Attribute (using the 0 value substitution and other rules) plus a secondary AF Attribute that will contain the quality / confidence factor of the value. That way you have better control over how to use the data, and your users/applications can choose to use the raw or cleansed data. Plus it is all configurable via AF & Templates.

          2 of 2 people found this helpful