pandas-dev/pandas · error · TypeError

Array with ndim > 2 is not supported.

Error message

Array with ndim > 2 is not supported.

What it means

Raised by pandas.core.algorithms.rank when the input array has more than 2 dimensions. rank only implements 1-D (rank_1d) and 2-D (rank_2d) paths; a 3-D+ array (e.g. a stacked tensor) has no supported ranking axis mapping.

Source

Thrown at pandas/core/algorithms.py:1254

            ties_method=method,
            ascending=ascending,
            na_option=na_option,
            pct=pct,
            mask=mask,
        )
    elif values.ndim == 2:
        assert mask is None
        ranks = algos.rank_2d(
            values,
            axis=axis,
            is_datetimelike=is_datetimelike,
            ties_method=method,
            ascending=ascending,
            na_option=na_option,
            pct=pct,
        )
    else:
        raise TypeError("Array with ndim > 2 is not supported.")

    return ranks


def is_monotonic(values: ArrayLike) -> tuple[bool, bool, bool]:
    """
    Determine whether values are monotonic increasing/decreasing.

    Parameters
    ----------
    values : np.ndarray or ExtensionArray

    Returns
    -------
    tuple[bool, bool, bool]
        (is_monotonic_increasing, is_monotonic_decreasing, is_strict_monotonic)

    Raises

View on GitHub (pinned to 71959b8cb9)

Solutions

1. Reduce to 1-D or 2-D before ranking (reshape, stack, or loop over the extra dimension).
2. Rank a 2-D slice (arr[:, :, 0]) and iterate over the third axis.
3. Use DataFrame.rank on a proper 2-D frame instead of a 3-D array.

Example fix

# before
arr = np.arange(24).reshape(2, 3, 4)
pd.core.algorithms.rank(arr)
# after
ranks = np.empty_like(arr, dtype=float)
for k in range(arr.shape[2]):
    ranks[:, :, k] = pd.core.algorithms.rank(arr[:, :, k])

import numpy as np, pandas as pd

def rank_any(arr, **kw):
    arr = np.asarray(arr)
    if arr.ndim > 2:
        out = np.empty(arr.shape, dtype=float)
        for idx in np.ndindex(arr.shape[2:]):
            full = (slice(None), slice(None)) + idx
            out[full] = pd.core.algorithms.rank(arr[full], **kw)
        return out
    return pd.core.algorithms.rank(arr, **kw)

import numpy as np

def is_rankable(arr) -> bool:
    return np.asarray(arr).ndim <= 2

When it happens

Trigger: Calling rank on a 3-D numpy array, or constructing a DataFrame/array whose underlying values are 3-D and routing it through the rank algorithm; reshaping data into ndim>2 before ranking.

Common situations: Multi-indexed/panel-like data flattened into a 3-D ndarray; passing raw np.ndarray of shape (n, m, k) to rank; custom ExtensionArray whose _values is 3-D.

Related errors

• Need to pass bool-like values
• values must be a 1D list-like
• mask must be a 1D list-like
• No such keys(s): {pat!r}
• {k} is not a valid identifier

AI-assisted analysis of pandas-dev/pandas@71959b8cb9 (2026-08-07). Data as JSON: /api/errors/d8af9961b9723792. Report an issue: GitHub.