pandas-dev/pandas · critical · ImportError

Please upgrade numpy to >=

Error message

Please upgrade numpy to >= {_min_numpy_ver} to use this pandas version.
Your numpy version is {_np_version}.

What it means

Raised at pandas import time (in pandas.compat.numpy) when the installed numpy is older than the floor pandas was built against (`_min_numpy_ver`, currently '2.0.2'). This is a hard import-time gate: pandas will not finish importing, so any `import pandas` fails before user code runs. The message states both the required minimum and the detected version.

Solutions

1. Upgrade numpy: `pip install -U 'numpy>=2.0.2'` or `conda install 'numpy>=2.0.2'`.
2. If you must keep the old numpy, install an older pandas whose `_min_numpy_ver` is satisfied (e.g. pandas 2.2.x).
3. Resolve dependency conflicts with `pip check` and a clean venv.

Example fix

// before
pip install numpy==1.26
import pandas as pd  # fails

// after
pip install 'numpy>=2.0.2'
import pandas as pd

import numpy as np
from packaging.version import Version
assert Version(np.__version__) >= Version('2.0.2')

When it happens

Trigger: `import pandas as pd` in an environment where numpy is pinned below 2.0.2 (e.g. a legacy environment with numpy 1.26); CI matrix accidentally testing old numpy with new pandas.

Common situations: Pinning numpy for another library (e.g. an old tensorflow) below pandas' floor; fresh conda env that resolved an old numpy; mixed pip+conda installs.

Related errors

• Pandas requires version
• Can't determine version for
• Cannot apply ufunc to mixed DataFrame and Series inputs.
• Cannot modify read-only array
• expected dimension <= 1 data

AI-assisted analysis of pandas-dev/pandas@3b7651241d (2026-08-11). Data as JSON: /api/errors/9492f898b3f48768. Report an issue: GitHub.

Appendix: source

Thrown at pandas/compat/numpy/__init__.py:19

"""support numpy compatibility across versions"""

import numpy as np

from pandas.util.version import Version

# numpy versioning
_np_version = np.__version__
_nlv = Version(_np_version)
np_version_gt2_2 = _nlv >= Version("2.2.0")
np_version_gt2_3 = _nlv >= Version("2.3.0")
np_version_gt2_5 = _nlv >= Version("2.5.0")
np_version_gt2_6 = _nlv >= Version("2.6.0.dev0")
is_numpy_dev = _nlv.dev is not None
_min_numpy_ver = "2.0.2"


if _nlv < Version(_min_numpy_ver):
    raise ImportError(
        f"Please upgrade numpy to >= {_min_numpy_ver} to use this pandas version.\n"
        f"Your numpy version is {_np_version}."
    )


__all__ = [
    "_np_version",
    "is_numpy_dev",
]

View on GitHub (pinned to 3b7651241d)