input.py

from params import fp_params, data_params, procedure_params, logging_config
import sys
from src import (
    lc,
    multiprocessing_logging,
    log_inputs,
    construct_fingerprint,
    prepare_data,
    run_single_model,
    run_multiprocessing,
    single_model_mp,
)

# partition given train/test indices
p = int(sys.argv[1])

# configure logger
lc.dictConfig(logging_config)
multiprocessing_logging.install_mp_handler()

# log inputs
log_inputs()

if __name__ == "__main__":

    # construct fingerprint
    if procedure_params["construct_fp"]:
        construct_fingerprint(**fp_params)

    # machine learning
    if (
        procedure_params["train"]
        or procedure_params["predict"]
        or procedure_params["hyperparametrize"]
    ):

        # data
        X, y = prepare_data(**data_params)

        # single model
        if not procedure_params["multiprocessing"]:
            run_single_model(X=X, y=y, p=p)

        # several models across processors
        if procedure_params["multiprocessing"]:
            run_multiprocessing(single_model=single_model_mp, X=X, y=y)