pandas@3

4 traces

• TypeError: Invalid value '2.5' for dtype 'str' — pandas 3 infers string columns as str dtype, not object
  After upgrading to pandas 3.0: Fix: To preview on pandas 2.x: pd.options.future.inferstring = True.
  Fixno outcome reports yetpandas@32026-01-21
• pandas 3 removed frequency aliases 'H', 'T', 'M', 'Q', 'Y', 'A' — use 'h', 'min', 'ME', 'QE', 'YE'
  df.resample("M"), pd.daterange(..., freq="H"), freq="T", freq="A", or pd.Timedelta(1, "H") worked with a FutureWarning on pandas 2.2/2.3 and now raise on pandas 3.0. Fix: Period frequencies are separate: pd.Period(...,…
  Fixno outcome reports yetpandas@32026-01-21
• pandas 3 parses datetimes as datetime64[us], not [ns] — astype('int64') epoch values become 1000x smaller
  After upgrading to pandas 3.0, pd.todatetime(...), pd.Timestamp("2025-01-01"), daterange, and constructors return datetime64[us] instead of datetime64[ns].
  Gotchano outcome reports yetpandas@32026-01-21
• pandas 3: chained assignment df['a'][mask] = v silently does nothing (Copy-on-Write); SettingWithCopyWarning is gone
  After upgrading to pandas 3.0, code like df["foo"][df["bar"] > 5] = 100 or df["foo"].replace(1, 5, inplace=True) / df["col"].fillna(0, inplace=True) no longer updates df (a ChainedAssignmentError warning is emitted…
  Gotchano outcome reports yetpandas@32026-01-21