pandas-dev/pandas · error · NotImplementedError

interpolate is not implemented for dtype=

Error message

interpolate is not implemented for dtype={self.dtype}

What it means

When interpolate falls back to the numpy-based path (non-fast-path method/limit/area settings), only float (kind 'f') and integer (kind 'iu') dtypes have a to_numpy conversion defined. Any other numeric-ish dtype (e.g. decimal[pyarrow]) hits the else branch and raises NotImplementedError. This complements error 154 by catching exotic numeric dtypes that passed _is_numeric but lack a numpy view path.

Solutions

1. Cast to float64 before interpolating: s.astype('float64').interpolate(...).
2. Use the default fast path (method='linear', no limit/limit_area) which stays inside pyarrow for numeric types.
3. Fall back to .ffill()/.bfill() for non-float numeric dtypes where precision loss is unacceptable.

Example fix

// before
s = pd.Series([Decimal("1"), None, Decimal("3")], dtype="decimal128(8,2)[pyarrow]")
s.interpolate(method="linear")
// after
s.astype("float64").interpolate(method="linear")

def safe_interpolate_fallback(s, **kw):
    if s.dtype.kind not in "iu":
        # only float/int have a numpy fallback path
        s = s.astype("float64")
    return s.interpolate(**kw)

def dtype_has_numpy_view(dtype) -> bool:
    return dtype.kind in {"f", "i", "u"}

try:
    s.interpolate(method="linear")
except NotImplementedError:
    s.astype("float64").interpolate(method="linear")

When it happens

Trigger: Calling .interpolate(method='slinear') or any non-default method on a decimal[pyarrow] or other non-float/int numeric ArrowExtensionArray.

Common situations: Decimal128[pyarrow] columns; future numeric pyarrow types that pandas recognizes as numeric but cannot convert cheaply.

Related errors

• Cannot interpolate with
• Converting strings to
• {pa_type}
• does not support reshape as backed by a 1D…
• ambiguous is not supported.

AI-assisted analysis of pandas-dev/pandas@3b7651241d (2026-08-11). Data as JSON: /api/errors/d8f1b1bbb16aa82d. Report an issue: GitHub.

Appendix: source

Thrown at pandas/core/arrays/arrow/array.py:3198

            method == "linear"
            and limit_area is None
            and limit is None
            and limit_direction == "forward"
        ):
            values = self._pa_array.combine_chunks()
            na_value = pa.array([None], type=values.type)
            y_diff_2 = pc.fill_null_backward(pc.pairwise_diff_checked(values, period=2))
            prev_values = pa.concat_arrays([na_value, values[:-2], na_value])
            interps = pc.add_checked(prev_values, pc.divide_checked(y_diff_2, 2))
            return self._from_pyarrow_array(pc.coalesce(self._pa_array, interps))

        mask = self.isna()
        if self.dtype.kind == "f":
            data = self._pa_array.to_numpy()
        elif self.dtype.kind in "iu":
            data = self.to_numpy(dtype="f8", na_value=0.0)
        else:
            raise NotImplementedError(
                f"interpolate is not implemented for dtype={self.dtype}"
            )

        missing.interpolate_2d_inplace(
            data,
            method=method,
            axis=0,
            index=index,
            limit=limit,
            limit_direction=limit_direction,
            limit_area=limit_area,
            mask=mask,
            **kwargs,
        )
        return self._from_pyarrow_array(self._box_pa_array(pa.array(data, mask=mask)))

    @classmethod
    def _if_else(

View on GitHub (pinned to 3b7651241d)