"""
Data loader for logistics-disruption-archive
"""

import pandas as pd
from pathlib import Path


def load_data(split: str = "train") -> pd.DataFrame:
    """
    Load the dataset.
    
    Args:
        split: Data split ("train" by default)
    
    Returns:
        DataFrame with the loaded data
    """
    data_dir = Path(__file__).parent / "data"
    parquet_path = data_dir / f"{split}.parquet"
    
    if not parquet_path.exists():
        raise FileNotFoundError(f"Data file not found: {parquet_path}")
    
    return pd.read_parquet(parquet_path)


def get_columns() -> list:
    """Return list of column names."""
    return ['supplier_diversity', 'delivery_consistency', 'inventory_buffer', 'chain_performance']


if __name__ == "__main__":
    df = load_data()
    print(f"Loaded {len(df)} rows")
    print(df.head())