--feat_archetype_similarity

Switch

--feat_archetype_similarity ARCHETYPE_FEAT_TABLE

Description

Creates a feature table of each group's similarity to a set of "archetype" groups, using the groups in ARCHETYPE_FEAT_TABLE as the archetypes to compare against. Aliased as --archtype_similarity.

Argument and Default Value

The name of a feature table whose groups serve as archetypes. Default: None.

Details

Known issue — crashes whenever group ids are numeric. FeatureRefiner.createArchetypeFeats() calls dlac.warn("calculating for: " + archName) where archName is one of the archetype table's raw group ids. Verified: with an integer -g field (e.g. user_id), this fails immediately with TypeError: can only concatenate str (not "int") to str as soon as it starts iterating archetypes — i.e. on essentially any normal numeric-group-id usage. Only string-typed group ids would avoid this particular crash; not otherwise verified end-to-end.

Other Switches

Required Switches:

• -d, -g, -t

• -f (the feature table to compare against the archetypes)

Example Commands

Not usable as of this writing with numeric group ids — see Known issue above:

dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$cat_met_a30_2000_cp_w$msgs$user_id$16to16' \
--feat_archetype_similarity 'feat$cat_met_a30_2000_cp_w$msgs$user_id$16to16'