Dependencies sample

Edit Subject

Roll ups are usually used for splitting of profiling results by some criteria.

If you have in input data records from more than one input system and ID of that system is present in every record, you can add a new roll up:
- Name “By system” and
- Expression “”.



After profiling run you will see additional roll-up windows:



Dependancy analysis is usually used to find possible denormalized enumerated values in the data. For example whether is there a code for an attribute and its decoded value from reference table (system id and system name, city code and city name).
You can add a dependancy analysis in a Profiler step on dependancy analysis tab. You specify a determinant (usually the column with the code) and one or more dependants (columns with possibly related values). With dependant, you specify possible threshold, which is ratio between number of a dependant values which depends on a determinant value and number of those that do not (the record with dependant value different then other dependant values for the determinant).


On the output you may see the number of those, which are trivially dependant (there is only 1 record with such a determinant), non-trivially dependant and number of Violations (records, where determinant does not determine a dependant value, there are multiple dependant values).



Feel free to ask any further questions. I understand that dependancy analysis is one of the most complex analysis to do with DQ Analyzer.