Method shuffle

dask/dataframe/dask_expr/_collection.py:844–954

Rearrange DataFrame into new partitions Uses hashing of `on` to map rows to output partitions. After this operation, rows with the same value of `on` will be in the same partition. Parameters ---------- on : str, list of str, or Series, Index, or Dat

(
        self,
        on: str | list | no_default = no_default,  # type: ignore[valid-type]
        ignore_index: bool = False,
        npartitions: int | None = None,
        shuffle_method: str | None = None,
        on_index: bool = False,
        force: bool = False,
        **options,
    )

        return self.partitions[n]

    def shuffle(
        self,
        on: str | list | no_default = no_default,  # type: ignore[valid-type]
        ignore_index: bool = False,
        npartitions: int | None = None,
        shuffle_method: str | None = None,
        on_index: bool = False,
        force: bool = False,
        **options,
    ):
        """Rearrange DataFrame into new partitions

        Uses hashing of `on` to map rows to output partitions. After this
        operation, rows with the same value of `on` will be in the same
        partition.

        Parameters
        ----------
        on : str, list of str, or Series, Index, or DataFrame
            Column names to shuffle by.
        ignore_index : optional
            Whether to ignore the index. Default is ``False``.
        npartitions : optional
            Number of output partitions. The partition count will
            be preserved by default.
        shuffle_method : optional
            Desired shuffle method. Default chosen at optimization time.
        on_index : bool, default False
            Whether to shuffle on the index. Mutually exclusive with 'on'.
            Set this to ``True`` if 'on' is not provided.
        force : bool, default False
            This forces the optimizer to keep the shuffle even if the final
            expression could be further simplified.
        **options : optional
            Algorithm-specific options.

        Notes
        -----
        This does not preserve a meaningful index/partitioning scheme. This
        is not deterministic if done in parallel.

        Examples
        --------
        >>> df = df.shuffle(df.columns[0])  # doctest: +SKIP
        """
        if on is no_default and not on_index:  # type: ignore[unreachable]
            raise TypeError(
                "Must shuffle on either columns or the index; currently shuffling on "
                "neither. Pass column(s) to 'on' or set 'on_index' to True."
            )
        elif on is not no_default and on_index:
            raise TypeError(
                "Cannot shuffle on both columns and the index. Do not pass column(s) "
                "to 'on' or set 'on_index' to False."
            )

        # Preserve partition count by default
        npartitions = npartitions or self.npartitions