--pickle

Switch

--pickle

Description

Generates a pickle file of the correlation-matrix-style output, alongside the html/csv output.

Argument and Default Value

No argument. Boolean switch, default: False.

Details

Not to be confused with the pickled-model workflow (tut_pickle_build/tut_pickle_apply) — that uses entirely different switches: --save_model (an argparse-abbreviation of --save_models/--save), --load_model (likewise for --load_models/--load), and --picklefile. Those work correctly and are unrelated to this page. --pickle (dest='pickle', no relation to args.picklefile) is a much narrower, separate switch.

Verified scope — narrower than it looks: --pickle is only checked inside the --cca output path (if args.rmatrix: ... if args.pickle: ... under the CCA branch of dlatkInterface.py) — confirmed by grepping every args.pickle reference in the codebase; there is exactly one, right there. It has no effect when used with the ordinary --correlate workflow — confirmed by running --correlate --csv --rmatrix --pickle and observing only .csv/.html files were written, no .pickle. If you need the general --correlate output as a Python object, there is currently no built-in pickle switch for that path; use the csv output instead.

When combined with --cca and --rmatrix, the code path would write <outputFile>.feat.pickle and <outputFile>.outcome.pickle versions of the feature/outcome correlation matrices — but this is unreachable as of this writing, since --cca itself always crashes before producing any results (a scoping bug unrelated to this switch — see the Known issue on --cca).

Other Switches

Required Switches:

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 gender --cca --rmatrix --pickle --output_name ./cca_out