data_set.py 文件源码

def load(path: Path) -> DataSet:
    """
    Loads a data set from the specified NetCDF4 file.

    Parameters
    ----------
    path: pathlib.Path
        Path to the file which should be loaded.

    Returns
    -------
    DataSet
        The data set loaded from the specified file
    """
    log = logging.getLogger(__name__)
    log.info("loading data set from %s", path)

    data = xr.open_dataset(str(path))  # type: xr.Dataset

    # restore data types
    data[_DataVar.FILENAME] = data[_DataVar.FILENAME].astype(np.object).fillna(None)
    data[_DataVar.CHUNK_NR] = data[_DataVar.CHUNK_NR].astype(np.object).fillna(None)
    data[_DataVar.CV_FOLDS] = data[_DataVar.CV_FOLDS].astype(np.object).fillna(None)
    data[_DataVar.PARTITION] = data[_DataVar.PARTITION].astype(np.object).fillna(None)
    data[_DataVar.LABEL_NOMINAL] = data[_DataVar.LABEL_NOMINAL].astype(np.object).fillna(None)
    data[_DataVar.LABEL_NUMERIC] = data[_DataVar.LABEL_NUMERIC].astype(np.object)
    data[_DataVar.FEATURES] = data[_DataVar.FEATURES].astype(np.float32)

    return DataSet(data=data,
                   mutable=False)