--weighted_sample

Switch

--weighted_sample FIELD

Description

A field in --outcome_table to use as per-group sample weights during regression training (e.g. survey/demographic weights).

Argument and Default Value

An outcome-table column name. Default: '' (dlac.DEF_WEIGHTS, no weighting).

Details

The named field is automatically appended to the requested outcomes (args.outcomefields.append(args.weightedsample)), then pulled back out and used as sample_weight when fitting the underlying scikit-learn regressor in RegressionPredictor.train() — every group's error contributes to the loss in proportion to its weight instead of equally.

Other Switches

Required Switches:

• One of the regression training switches, e.g. --train_regression, --nfold_test_regression

Example Commands

dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$1gram$msgs$user_id$16to16' \
--outcome_table blog_outcomes --outcomes age --weighted_sample survey_weight --combo_test_regression