The pandas team is pleased to announce the release of pandas 1.4.0.

You can install it via pip install pandas or mamba install pandas.

Full release notes are available at: https://pandas.pydata.org/pandas-docs/version/1.4/whatsnew/v1.4.0.html

Work has started to support non-nanosecond resultion dates in pandas.

Currently, pandas can represent dates around the year interval 1680 ~ 2260. An exception is raised if dates outside of this range are used.

When the new changes are released (probably at the end of this year), pandas will be able to represent billions of years from now into the past or the future, with a precision of seconds.

We just released pandas 1.4.2 a bug fix release of the 1.4 series. We recommend all users using pandas 1.4 to upgrade.

It can be installed via pip and from conda-forge.

Full release notes available here.

pandas is now in Open Collective, where you can see the project finances. Expenses and more details are still being added, but the main sources of income of the project are already available.

You will also find in Open Collective the rest of NumFOCUS supported projects, such as Jupyter, SciPy, etc.

The book "Python for Data Analysis", by original pandas creator Wes McKinney, has an open access version of it's 3rd edition, with all chapters.

https://wesmckinney.com/book/

Bodo has just become a pandas and NumFOCUS financial supporter. Bodo is a just in time compiler that allows executing pandas code in a parallel way without running the pandas underlying code based on NumPy. This should help run pandas code faster and at scale. Bodo funds will be used to improve pandas API, and will contribute to pandas general maintenance and to the wider community.

pandas is adopting PDEPs, similar to PEPs in Python and NEPs in NumPy. The formal proposal process should make discussions more efficients, and future plans clearer and more visible. First PDEP about PDEPs workflow has been proposed and is accepting (and welcoming) feedback:

https://github.com/pandas-dev/pandas/pull/47444

There is now a browser-based interactive terminal in the pandas website: https://pandas.pydata.org/getting_started.html

The Python interpreter and dependencies are shipped as a WebAssembly binary, and it doesn't use a backend, everything you run is executed locally in the browser process. The only limitation is that the browser runs in it a sandbox, and I/O is limited. For example, downloading files from other websites is not allowed due to cross-domain (CORS) limitations.

pandas-stubs 1.4.2.220622 has been released. This is the first official version of pandas-stubs by the pandas team, after merging the two third-party projects Microsoft python-type-stubs and VirtusLabs pandas_stubs.

pandas-stubs is useful to have type checking in your pandas projects. By installing pandas-stubs you can validate that types in your code are consistent. You can see an example here: https://github.com/pandas-dev/pandas-stubs#usage

You can install pandas-stubs via pip and conda-forge.

We are pleased to announce the release of pandas v1.4.3.

This is a patch release in the 1.4.x series and includes some regression fixes and bug fixes. We recommend that all users in the 1.4.x series upgrade to this version.

See the release notes for a list of all the changes.

The release can be installed from PyPI

    python -m pip install --upgrade pandas==1.4.3

Or from conda-forge

    conda install -c conda-forge pandas==1.4.3

pandas core dev Jeff Reback gave a talk on the past, present and future of pandas: https://m.youtube.com/watch?v=7JHqxODJG9k

pandas added some new sponsors in the last few months. Thanks Voltron Data, Quansight Labs, NVIDIA and Bodo for supporting pandas development. You can see the full list of sponsors in our home page: https://pandas.pydata.org/

PDEP-1 has been accepted, being the first pandas enhancement proposal, and setting the guidelines fo discussions about the pandas future. This will not only make discussions easier, but it will also bring clarity and transparency to the project plans.

We just updated our Team page to show more clearly the active maintainers who are currently working on pandas (in a broad sense, 22 maintainers).

We want to thank the maintainers who are not active anymore, but who helped build what pandas is today: Wouter Overmeire, Skipper Seabold, Jeff Tratner, Stephan Hoyer, chris-b1, Sinhrks, Phillip Cloud, Pietro Battiston, Jeremy Schendel, Kaiqi Dong, and Daniel Saxton.