def main()

in bayesmark/experiment_aggregate.py [0:0]

• 23 lines of code
• 6 McCabe index (conditional complexity)

def main():
    """See README for instructions on calling aggregate.
    """
    description = "Aggregate study results across functions and optimizers"
    args = parse_args(agg_parser(description))

    logger.setLevel(logging.INFO)  # Note this is the module-wide logger
    if args[CmdArgs.verbose]:
        logger.addHandler(logging.StreamHandler())

    # Get list of UUIDs
    uuid_list = XRSerializer.get_uuids(args[CmdArgs.db_root], db=args[CmdArgs.db], key=cc.EVAL)
    uuid_list_ = XRSerializer.get_uuids(args[CmdArgs.db_root], db=args[CmdArgs.db], key=cc.TIME)
    assert uuid_list == uuid_list_, "UUID list does not match between time and eval results"
    uuid_list_ = XRSerializer.get_uuids(args[CmdArgs.db_root], db=args[CmdArgs.db], key=cc.SUGGEST_LOG)
    assert uuid_list == uuid_list_, "UUID list does not match between suggest log and eval results"

    # Get iterator of all experiment data dumps, load in and process, and concat
    data_G = load_experiments(uuid_list, args[CmdArgs.db_root], args[CmdArgs.db])
    all_perf, all_time, all_suggest, all_sigs = concat_experiments(data_G, ravel=args[CmdArgs.ravel])

    # Check the concat signatures make are coherent
    sig_errs, signatures_median = analyze_signatures(all_sigs)
    logger.info("Signature errors:\n%s" % sig_errs.to_string())
    print(json.dumps({"exp-agg sig errors": sig_errs.T.to_dict()}))

    # Dump and save it all out
    logger.info("saving")
    meta = {"args": serializable_dict(args), "signature": signatures_median}
    XRSerializer.save_derived(all_perf, meta, args[CmdArgs.db_root], db=args[CmdArgs.db], key=EVAL_RESULTS)
    XRSerializer.save_derived(all_time, meta, args[CmdArgs.db_root], db=args[CmdArgs.db], key=TIME_RESULTS)
    for test_case, ds in all_suggest.items():
        XRSerializer.save_derived(ds, meta, args[CmdArgs.db_root], db=args[CmdArgs.db], key=test_case)

    logger.info("done")