--train_bootstraps

Switch

--train_bootstraps N

Description

Number of bootstrap resamples to train, when training a regression model with --train_regression.

Argument and Default Value

An integer. Default: None (no bootstrapping — a single model is trained).

Details

Verified end-to-end: with --train_bootstraps 3, ran the training pipeline 3 times (per sample size — see --train_bootstraps_ns), printing Sample Size: <N>, bs resample num: 1/2/3 for each. Useful for estimating model-performance variance under repeated resampling, e.g. for learning curves.

Other Switches

Required Switches:

• --train_regression

Optional Switches:

• --train_bootstraps_ns

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 --train_regression --train_bootstraps 3 \
--train_bootstraps_ns 500 1000 --save_model --picklefile ./boot_test.pickle