scripts/convert_cmapss_to_parquet.py

#!/usr/bin/env python3
"""Convert NASA C-MAPSS raw train text files to processed parquet."""

from __future__ import annotations

import argparse
from pathlib import Path

import pandas as pd


CMAPSS_COLUMNS = (
    "unit",
    "cycle",
    "op1",
    "op2",
    "op3",
    *(f"s{index}" for index in range(1, 22)),
)


def convert_train_file(input_path: Path, output_path: Path) -> pd.DataFrame:
    frame = pd.read_csv(
        input_path,
        sep=r"\s+",
        header=None,
        names=CMAPSS_COLUMNS,
    )
    max_cycle = frame.groupby("unit")["cycle"].transform("max")
    frame["rul"] = (max_cycle - frame["cycle"]).astype(int)
    output_path.parent.mkdir(parents=True, exist_ok=True)
    frame.to_parquet(output_path, index=False)
    return frame


def main() -> int:
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument(
        "--raw-dir",
        type=Path,
        default=Path("data/raw/cmapss"),
        help="Directory containing NASA train_FD00X.txt files.",
    )
    parser.add_argument(
        "--output-dir",
        type=Path,
        default=Path("data"),
        help="Directory where FD00X_processed.parquet files are written.",
    )
    parser.add_argument(
        "--datasets",
        nargs="+",
        default=["FD002", "FD004"],
        help="C-MAPSS dataset IDs to convert.",
    )
    args = parser.parse_args()

    for dataset_id in args.datasets:
        source = args.raw_dir / f"train_{dataset_id}.txt"
        target = args.output_dir / f"{dataset_id}_processed.parquet"
        frame = convert_train_file(source, target)
        print(f"{dataset_id}: wrote {target} ({len(frame)} rows)")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())