39 lines
1.4 KiB
Python
Executable File
39 lines
1.4 KiB
Python
Executable File
#!/usr/bin/env python3
|
|
"""Prepare vocab and normalization stats for HAI 21.03."""
|
|
|
|
import json
|
|
from pathlib import Path
|
|
from typing import Optional
|
|
|
|
from data_utils import compute_cont_stats, build_vocab, load_split
|
|
from platform_utils import safe_path, ensure_dir
|
|
|
|
BASE_DIR = Path(__file__).resolve().parent
|
|
REPO_DIR = BASE_DIR.parent.parent
|
|
DATA_PATH = REPO_DIR / "dataset" / "hai" / "hai-21.03" / "train1.csv.gz"
|
|
SPLIT_PATH = BASE_DIR / "feature_split.json"
|
|
OUT_STATS = BASE_DIR / "results" / "cont_stats.json"
|
|
OUT_VOCAB = BASE_DIR / "results" / "disc_vocab.json"
|
|
|
|
|
|
def main(max_rows: Optional[int] = None):
|
|
split = load_split(safe_path(SPLIT_PATH))
|
|
time_col = split.get("time_column", "time")
|
|
cont_cols = [c for c in split["continuous"] if c != time_col]
|
|
disc_cols = [c for c in split["discrete"] if not c.startswith("attack") and c != time_col]
|
|
|
|
mean, std = compute_cont_stats(safe_path(DATA_PATH), cont_cols, max_rows=max_rows)
|
|
vocab = build_vocab(safe_path(DATA_PATH), disc_cols, max_rows=max_rows)
|
|
|
|
ensure_dir(OUT_STATS.parent)
|
|
with open(safe_path(OUT_STATS), "w", encoding="utf-8") as f:
|
|
json.dump({"mean": mean, "std": std, "max_rows": max_rows}, f, indent=2)
|
|
|
|
with open(safe_path(OUT_VOCAB), "w", encoding="utf-8") as f:
|
|
json.dump({"vocab": vocab, "max_rows": max_rows}, f, indent=2)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
# Default: sample 50000 rows for speed. Set to None for full scan.
|
|
main(max_rows=50000)
|