Why could sktime 0.27.0 be in conflict with Pandas 2.2.1?

[matthewferler]

I have a Dockerfile that uses pip3 and requirements.txt to install several things including sktime and pandas. When I specify the combination of pandas==2.2.1 and sktime==0.27.0 there is a pip3 error. I don't understand why because the sktime docs indicate that Pandas 2 has been supported since version 0.19.1. The output from pip3 is as follows. Any help to understand this issue is greatly appreciated.

INFO: pip is looking at multiple versions of sktime to determine which version is compatible with other requirements. This could take a while.
ERROR: Cannot install -r /app/requirements.txt (line 5) and pandas==2.2.1 because these package versions have conflicting dependencies.

The conflict is caused by:
The user requested pandas==2.2.1
sktime 0.27.0 depends on pandas<2.2.0 and >=1.1

[fkiraly]

We manage an upper bound of pandas, since there are close dependencies on objects like the ForecastingHorizon on pandas indicese, and MINOR version releases of pandas, in the past, have repeatedly broken core functionality.

An update that makes sktime compatible with pandas 2.2.X will be available in sktime 0.28.0, scheduled for the next days.

Apologies, the pandas incompatibilities tend to be increasingly obscure, as they have to do with freq and multiples of exotic frequency strings and the like.

[fkiraly]

In case you are an appreciator of subtle horror, here is the respective fix: #6057

Generally, contributions that help us address various warnings from future deprecations early are much appreciated!

Also, the ForecastingHorizon will need a rework at some time in the future, that's also a mini-project where help is appreciated.

[matthewferler]

Thank you for the information. It's a great library. I look forward to 0.28.0.

[fkiraly]

FYI @matthewferler, sktime 0.28.0 has been released and is compatible with pandas 2.2.X. Hopefully, no further issues arise from the changes in how pandas handles freq.

[matthewferler]

I pulled sktime 0.28.0 into my environment along with pandas 2.2.1 successfully. Thank you for all of your hard work!

[fkiraly]

You're welcome!
(donations of course are also appreciated 😉 https://opencollective.com/sktime )