pandas-dev/pandas · error · TypeError

Cannot perform {how} with non-ordered Categorical

Error message

Cannot perform {how} with non-ordered Categorical

What it means

Raised by Categorical._groupby_op when the aggregation is order-based (min, max, rank, idxmin, idxmax) but the categorical dtype is unordered. Min/max/rank require a total ordering; an unordered categorical defines only a label set, so these operations are undefined. The comment in source notes this is raised as TypeError deliberately to avoid a slower per-group path that would also fail.

Source

Thrown at pandas/core/arrays/categorical.py:2876

        has_dropped_na: bool,
        min_count: int,
        ngroups: int,
        ids: npt.NDArray[np.intp],
        **kwargs,
    ):
        from pandas.core.groupby.ops import WrappedCythonOp

        kind = WrappedCythonOp.get_kind_from_how(how)
        op = WrappedCythonOp(how=how, kind=kind, has_dropped_na=has_dropped_na)

        dtype = self.dtype
        if how in ["sum", "prod", "cumsum", "cumprod", "skew", "kurt"]:
            raise TypeError(f"{dtype} type does not support {how} operations")
        if how in ["min", "max", "rank", "idxmin", "idxmax"] and not dtype.ordered:
            # raise TypeError instead of NotImplementedError to ensure we
            #  don't go down a group-by-group path, since in the empty-groups
            #  case that would fail to raise
            raise TypeError(f"Cannot perform {how} with non-ordered Categorical")
        if how not in [
            "rank",
            "any",
            "all",
            "first",
            "last",
            "min",
            "max",
            "idxmin",
            "idxmax",
        ]:
            if kind == "transform":
                raise TypeError(f"{dtype} type does not support {how} operations")
            raise TypeError(f"{dtype} dtype does not support aggregation '{how}'")

        result_mask = None
        mask = self.isna()
        if how == "rank":

View on GitHub (pinned to 71959b8cb9)

Solutions

1. Make the column ordered: df['cat'] = df['cat'].cat.as_ordered() (or pd.Categorical(..., ordered=True) with explicit categories).
2. Use .first()/.last() instead of min/max if ordering is not meaningful.
3. Cast to the underlying value dtype and aggregate there.

Example fix

// before
df.groupby('g')['cat'].min()  # TypeError: Cannot perform min with non-ordered Categorical

// after
df['cat'] = df['cat'].cat.as_ordered()
df.groupby('g')['cat'].min()

ORDER_HOW = {'min','max','rank','idxmin','idxmax'}
def ensure_ordered_for_how(s, how):
    import pandas as pd
    if isinstance(s.dtype, pd.CategoricalDtype) and how in ORDER_HOW and not s.cat.ordered:
        return s.cat.as_ordered()
    return s

import pandas as pd
from typing import Any

def can_perform_order_op(obj: Any, how: str) -> bool:
    dt = getattr(obj, 'dtype', None)
    if isinstance(dt, pd.CategoricalDtype):
        return dt.ordered or how not in {'min','max','rank','idxmin','idxmax'}
    return True

try:
    df.groupby('g')['cat'].min()
except TypeError as e:
    if 'non-ordered Categorical' in str(e):
        df['cat'] = df['cat'].cat.as_ordered()
    else:
        raise

When it happens

Trigger: df.groupby('g')['cat'].min(), .max(), .rank(), .idxmin(), or .idxmax() on an unordered category column.

Common situations: A 'low/med/high' or 'cold/warm/hot' style column created without ordered=True; groupby pipelines that compute extrema or ranks on every categorical column.

Related errors

• 'values' is not ordered, please explicitly specify the categ
• Categorical is not ordered for operation {op} you can use .a
• {dtype} type does not support {how} operations
• numpy operations are not valid with groupby. Use .groupby(..
• dtype '{self.dtype}' does not support operation '{how}'

AI-assisted analysis of pandas-dev/pandas@71959b8cb9 (2026-08-07). Data as JSON: /api/errors/2b2861f2abc1a5ec. Report an issue: GitHub.