--mask_table

Switch

--mask_table TABLE

Description

A table indicating which groups belong to which "mask" (subset), for running --ttest_feat_tables separately within each mask rather than over all groups at once.

Argument and Default Value

A table name. Default: None (no masking; one ttest run over all shared groups).

Details

Expected format: one column matching your -g group field, and one additional column per mask, with a 1/0 (or truthy) value marking whether each group belongs to that mask. A separate paired t-test is run per mask, intersected with the groups passing --group_freq_thresh in both feature tables.

Likely SQLite issue (not run): FeatureGetter.ttestWithOtherFG() reads the mask table using mm.getTableColumnNameList() and mm.executeGetList() — calls straight into mysqlmethods, bypassing the DataEngine/QueryBuilder abstraction that gives the rest of DLATK its SQLite support. This suggests --mask_table (and possibly all of --ttest_feat_tables's masking path) only works against a real MySQL connection. Not independently reproduced in SQLite mode — flagged from reading the source, not from a failing run.

Other Switches

Required Switches:

• --ttest_feat_tables

• --group_freq_thresh

Example Commands

dlatkInterface.py -d dla_tutorial -t msgs -c user_id \
-f 'feat$1gram$msgs_before$user_id' 'feat$1gram$msgs_after$user_id' \
--ttest_feat_tables --group_freq_thresh 100 --mask_table region_masks