I'm in the middle of wrapping up the edits for Effective Pandas 3rd Edition. (I also wrote a Polars book and just wrapped up a weeklong training session on pandas this week.)

Pandas is not perfect, it has a bunch of warts. But it is good enough for most. (And many of those folks are using Excel or tableau or power bi... These were the types I was training this week).

If you have medium data, migrating from pyarrow backed pandas to duck or Polars is trivial.