INNER CODE UNIT · Python

load_sample

google/fully-homomorphic-encryption · demos/cc_fraud/cleartext/evaluate_cleartext.py:20

def load_sample(
    data_path: str, feature_cols: list[str], sample_idx: int
) -> tuple[np.ndarray, int]:
  """Loads a single sample from the dataset."""
  resolved_data_path = resolve_path(data_path)
  if not os.path.exists(resolved_data_path):
    raise FileNotFoundError(
        f"Data file not found at {data_path} (resolved: {resolved_data_path})"
    )

  if data_path.endswith(".parquet"):
    df = pd.read_parquet(resolved_data_path)
    X = df[feature_cols].values.astype(np.float32)
    y = df["is_fraud"].values.astype(np.int64)
    # Retrieve the test split (same random seed/stratify as train.py)
    _, X_test, _, y_test = train_test_split(
        X, y, test_size=0.2, random_state=42, stratify=y
    )

View source record →