pandas > 1.5 reason? - Githubissues

raphaelvallat / pingouin

Statistical package in Python based on Pandas

https://pingouin-stats.org/

GNU General Public License v3.0

1.63k stars 139 forks source link

pandas > 1.5 reason? #416

Closed info-rchitect closed 6 months ago

info-rchitect commented 7 months ago

Hi,

Was curious if there was a specific reason to fix pandas > 1.5? The main reason being is many commercial platforms, for example Dataiku, are very slow to upgrade their infrastructure for pandas. Could > 1.3 work?

thx

raphaelvallat commented 6 months ago

Hey,

Pandas 1.5 introduced the numeric_only parameter in many functions

Pandas 2+ changed the default value of numeric_only=False for most functions, which led to many user-reported errors. To avoid this, we added numeric_only=True to all the relevant functions, hence the dependency for pandas 1.5.

I'm not in favour of removing it because all these errors would then resurface in individual users.

Thanks Raphael

info-rchitect commented 6 months ago

@raphaelvallat Thanks for taking the time to explain this. We use Dataiku as a primary analytics tool and it is currently limited to pandas 1.35 for Dataiku versions < 12.5.1. Once we convert to that we can use pandas 1.5.