Extension pandas dtype and array for physical units

This python package extends the popular data science library pandas with the physical units of the astrophysics library astropy. The package is compatible with modern pandas, although some features are only working with pandas version 3.0 or an upcoming pandas release.

History

Originally created for PyCon CZ 2019 (and later BI Fórum Budapest / PyData Cambridge the same year) by Jan Pipek. Updated to be compatible with modern pandas since 2025 by Julian Harbeck during the RACCOON project at Technische Universität Berlin.

Installation

pip install pandas-units-extensions

For development:

pip install -e .

Examples

>>> import pandas as pd
>>> import pandas_units_extension as pue

>>> temps = pd.DataFrame({
        "city": ["Prague", "Kathmandu", "Catania", "Boston"],
        "temperature": pd.Series([20, 22, 31, 16], dtype="unit[deg_C]")
    })
>>> temps["temperature"].units.to("deg_F")
0    68.0 deg_F
1    71.6 deg_F
2    87.8 deg_F
3    60.8 deg_F
Name: temperature, dtype: unit[deg_F]


>>> df = pd.DataFrame({
        "distance": pd.Series([10, 12, 22, 18], dtype="unit[km]"),
        "time": pd.Series([50, 60, 120, 108], dtype="unit[min]")
    })
>>> speed = df["distance"] / df["time"]
>>> speed.units.to_si()
0     3.333333333333334 m / s
1     3.333333333333334 m / s
2    3.0555555555555554 m / s
3    2.7777777777777777 m / s
dtype: unit[m / s]

See doc/units.ipynb for more.

Links

• https://www.astropy.org/
• https://pint.readthedocs.io/en/0.10.1/pint-pandas.html - Another library supporting units inside pandas.