pandas-dev/pandas · error · TypeError

'std' and 'sem' are not valid for PeriodDtype

Error message

'std' and 'sem' are not valid for PeriodDtype

What it means

Raised in _groupby_op after the cython reduction, when how is 'std' or 'sem' and self.dtype is PeriodDtype. Standard deviation of Periods is undefined because subtraction of Periods is not a Period operation, so even though std/sem pass the earlier allow-list they are explicitly rejected here for Period data only (DatetimeArray/TimedeltaArray return a Timedelta result instead).

Solutions

  1. Convert to timestamps first: s.dt.to_timestamp().groupby(g).std().
  2. If you want ordinal dispersion, cast to int64 via .view('i8') or .astype('int64') then compute std and interpret carefully.
  3. Switch from PeriodDtype to a DatetimeIndex/DatetimeArray if spread statistics are a core requirement.

Example fix

// before
s = pd.Series(pd.PeriodIndex(['2020-01','2020-02','2020-03'], freq='M'))
s.std()  # TypeError: 'std' and 'sem' are not valid for PeriodDtype

// after
s.dt.to_timestamp().std()
Defensive patterns

Strategy: validation

Validate before calling

def period_std(s, group):
    if isinstance(s.dtype, pd.PeriodDtype):
        return s.dt.to_timestamp().groupby(group).std()
    return s.groupby(group).std()

Type guard

def is_period_std_unsupported(dtype, how) -> bool:
    return isinstance(dtype, pd.PeriodDtype) and how in {"std","sem"}

Try / catch

try:
    out = s.groupby(g).std()
except TypeError as e:
    if "not valid for PeriodDtype" in str(e):
        out = s.dt.to_timestamp().groupby(g).std()
    else:
        raise

Prevention

When it happens

Trigger: groupby(...).std() or .sem() (and resample/rolling equivalents) on a Series/Index with PeriodDtype.

Common situations: Statistical summaries over monthly/quarterly Period columns; polymorphic .describe()-style calls that include std across mixed dtypes; assuming std behaves on Period like it does on Datetime.

Related errors

AI-assisted analysis of pandas-dev/pandas@3b7651241d (2026-08-11). Data as JSON: /api/errors/bce5cbfda5de70b6. Report an issue: GitHub.

Appendix: source

Thrown at pandas/core/arrays/datetimelike.py:1684

            min_count=min_count,
            ngroups=ngroups,
            comp_ids=ids,
            mask=None,
            **kwargs,
        )

        if op.how in op.cast_blocklist:
            # i.e. how in ["rank"], since other cast_blocklist methods don't go
            #  through cython_operation
            return res_values

        # We did a view to M8[ns] above, now we go the other direction
        assert res_values.dtype == "M8[ns]"
        if how in ["std", "sem"]:
            from pandas.core.arrays import TimedeltaArray

            if isinstance(self.dtype, PeriodDtype):
                raise TypeError("'std' and 'sem' are not valid for PeriodDtype")
            self = cast("DatetimeArray | TimedeltaArray", self)
            new_dtype = f"m8[{self.unit}]"
            res_values = res_values.view(new_dtype)
            return TimedeltaArray._simple_new(res_values, dtype=res_values.dtype)

        res_values = res_values.view(self._ndarray.dtype)
        return self._from_backing_data(res_values)

    def _groupby_quantile(
        self,
        *,
        qs: npt.NDArray[np.float64],
        interpolation: Literal["linear", "lower", "higher", "nearest", "midpoint"],
        ids: npt.NDArray[np.intp],
        ngroups: int,
        starts: npt.NDArray[np.int64],
        ends: npt.NDArray[np.int64],
    ) -> ArrayLike:

View on GitHub (pinned to 3b7651241d)