pandas-dev/pandas · error · NotImplementedError

axis other than 0 is not supported

Error message

axis other than 0 is not supported

What it means

Raised in `agg_or_apply_list_like` when the object's axis attribute is 1 (column-wise operation) for a list-like agg/apply path that only supports axis=0. List-like aggregation iterates columns, so the implementation deliberately rejects axis=1 to avoid surprising behavior.

Source

Thrown at pandas/core/apply.py:882

    def agg_or_apply_list_like(
        self, op_name: Literal["agg", "apply"]
    ) -> DataFrame | Series:
        obj = self.obj
        kwargs = self.kwargs

        if op_name == "apply":
            if isinstance(self, FrameApply):
                by_row = self.by_row

            elif isinstance(self, SeriesApply):
                by_row = "_compat" if self.by_row else False
            else:
                by_row = False
            kwargs = {**kwargs, "by_row": by_row}

        if getattr(obj, "axis", 0) == 1:
            raise NotImplementedError("axis other than 0 is not supported")

        if op_name == "agg" and isinstance(self, FrameApply):
            result = self._agg_list_like_frame_reductions()
            if result is not None:
                return result

        keys, results = self.compute_list_like(op_name, obj, kwargs)
        result = self.wrap_results_list_like(keys, results)
        return result

    def agg_or_apply_dict_like(
        self, op_name: Literal["agg", "apply"]
    ) -> DataFrame | Series:
        assert op_name in ["agg", "apply"]
        obj = self.obj

        kwargs = {}
        if op_name == "apply":

View on GitHub (pinned to 71959b8cb9)

Solutions

1. Drop axis=1 and let agg operate column-wise (axis=0), which is the only supported direction for list-like func.
2. Transpose the frame: `df.T.agg(['sum', 'mean']).T` if you genuinely need row-wise reduction.
3. Use `df.apply(func, axis=1)` with a single function that internally computes multiple stats, returning a Series.

Example fix

# before
df.agg(['sum', 'mean'], axis=1)
# after
df.T.agg(['sum', 'mean']).T

def safe_list_agg(df, funcs, axis=0):
    if axis == 1:
        raise NotImplementedError('list-like agg supports axis=0 only; use df.T')
    return df.agg(funcs, axis=axis)

def is_axis0_or_transposable(axis) -> bool:
    return axis == 0

try:
    df.agg(funcs, axis=1)
except NotImplementedError as e:
    if 'axis other than 0' in str(e):
        df.T.agg(funcs).T
    else:
        raise

When it happens

Trigger: `df.agg(['sum', 'mean'], axis=1)` — list-like func with axis=1. The check at apply.py:881 (`getattr(obj, 'axis', 0) == 1`) fires.

Common situations: Copy-pasting axis=1 from a single-function apply call into a list-based agg; assuming axis symmetry across all agg surfaces; refactoring from row-wise apply to list-agg without dropping axis.

Related errors

• cannot perform both aggregation and transformation operation
• Operation {func} does not support axis=1
• The 'numba' engine doesn't support list-like/dict likes of c
• nested renamer is not supported
• Label(s) {list(cols)} do not exist

AI-assisted analysis of pandas-dev/pandas@71959b8cb9 (2026-08-07). Data as JSON: /api/errors/04e3c986b2b90968. Report an issue: GitHub.