Under what conditions does a field qualify as a "category" field data type?
We are testing some reports against multiple data sets with the same structure (same fields) but different data points and believed that the "category" field data type applied to fields with a certain number of discrete text values (under 500?) - however, we are finding inconsistencies in the way these are interpreted with some fields being read as data type text and others data type category. We are querying from an ingres database if that helps.
Is there a way to guarantee the output data type value that then ultimately drives that type the filter is? Like some rules to follow? (Selectable options as opposed to text search boxes). We are finding that selecting category data type on the field is not enough if the data points underneath change.
Omniscope Classic uses a few heuristics to pick whether to classify a field as text or category.
- the number of unique values (excluding nulls) < 100 - and the average length of text values (excluding nulls) <= 50<br /> This is the defaults for an unconfigured field.
If you've configured a field as category, Omniscope will then force it to be text if a subsequent data update arrives where:
- the number of unique values (excluding nulls) < 1000 - and the average length of text values (excluding nulls) <= 1000<br /> This force effect can be disabled by unticking "Auto-convert text fields" advanced file setting.
So if you've configured a field in Field Organiser to be Text or Category, and you have that option unticked, it should stay text or category, regardless of new data refreshed.