--prefix_super_groups

Switch

--prefix_super_groups

Description

When stratifying cross-validation folds with --stratify_folds, treats a prefix of each group id as a "super group" to stratify by, instead of the full group id.

Argument and Default Value

No argument. Boolean switch, default: False.

Details

Passed as stratifyPrefixSuperGroups into RegressionPredictor.testControlCombos(). Useful when group ids are structured (e.g. a county FIPS code where the first few digits are a state), so folds are balanced across the higher-level grouping rather than the raw id.

Other Switches

Required Switches:

• --nfold_test_regression

• --stratify_folds

Example Commands

Using the packaged dla_tutorial dataset (see Differential Language Analysis (DLA) Tutorial):

dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$1to3gram$msgs$user_id$0_05' \
--outcome_table blog_outcomes --outcomes age --combo_test_regression \
--stratify_folds --prefix_super_groups