--predict_classifiers_to_outcome_table

Switch

--predict_classifiers_to_outcome_table NAME

Description

Predicts classes with a previously-trained classifier and writes them into a new outcome table, rather than a feature table (see --predict_classifiers_to_feats for the feature-table version).

Argument and Default Value

A name used to build the destination table (p_<model>$<name>). Default: None.

Details

Verified end-to-end. Requires a pre-trained, pre-loaded model — --load_models/--picklefile (trained earlier via --train_classifiers + --save_models). Without one, it fails with IndexError: list index out of range while trying to align features against an empty model. With a real loaded SVC model, produced a table (p_svc$<name>, one row per group with the predicted class, e.g. 743739|0).

Other Switches

Required Switches:

• -d, -g, -t

• -f

• --load_models, --picklefile

Example Commands

Using the packaged dla_tutorial dataset (see Building A Pickle Model and Applying A Pickle Model):

dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$1to3gram$msgs$user_id$0_05' \
--outcome_table blog_outcomes --outcomes is_student --train_classifiers --save_models \
--picklefile is_student.pickle

dlatkInterface.py -d dla_tutorial -t msgs -c user_id -f 'feat$1to3gram$msgs$user_id$0_05' \
--load_models --picklefile is_student.pickle --predict_classifiers_to_outcome_table pred_is_student