What’s new in 1.1.3 (October 5, 2020) — pandas 2.2.3 documentation (original) (raw)

These are the changes in pandas 1.1.3. See Release notes for a full changelog including other versions of pandas.

Enhancements#

Added support for new Python version#

pandas 1.1.3 now supports Python 3.9 (GH 36296).

Development Changes#

• The minimum version of Cython is now the most recent bug-fix version (0.29.21) (GH 36296).

Fixed regressions#

• Fixed regression in DataFrame.agg(), DataFrame.apply(), Series.agg(), and Series.apply() where internal suffix is exposed to the users when no relabelling is applied (GH 36189)
• Fixed regression in IntegerArray unary plus and minus operations raising a TypeError (GH 36063)
• Fixed regression when adding a timedelta_range() to a Timestamp raised a ValueError (GH 35897)
• Fixed regression in Series.__getitem__() incorrectly raising when the input was a tuple (GH 35534)
• Fixed regression in Series.__getitem__() incorrectly raising when the input was a frozenset (GH 35747)
• Fixed regression in modulo of Index, Series and DataFrame using numexpr using C not Python semantics (GH 36047, GH 36526)
• Fixed regression in read_excel() with engine="odf" caused UnboundLocalError in some cases where cells had nested child nodes (GH 36122, GH 35802)
• Fixed regression in DataFrame.replace() inconsistent replace when using a float in the replace method (GH 35376)
• Fixed regression in Series.loc() on a Series with a MultiIndex containing Timestamp raising InvalidIndexError (GH 35858)
• Fixed regression in DataFrame and Series comparisons between numeric arrays and strings (GH 35700, GH 36377)
• Fixed regression in DataFrame.apply() with raw=True and user-function returning string (GH 35940)
• Fixed regression when setting empty DataFrame column to a Series in preserving name of index in frame (GH 36527)
• Fixed regression in Period incorrect value for ordinal over the maximum timestamp (GH 36430)
• Fixed regression in read_table() raised ValueError when delim_whitespace was set to True (GH 35958)
• Fixed regression in Series.dt.normalize() when normalizing pre-epoch dates the result was shifted one day (GH 36294)

Bug fixes#

• Bug in read_spss() where passing a pathlib.Path as path would raise a TypeError (GH 33666)
• Bug in Series.str.startswith() and Series.str.endswith() with category dtype not propagating na parameter (GH 36241)
• Bug in Series constructor where integer overflow would occur for sufficiently large scalar inputs when an index was provided (GH 36291)
• Bug in DataFrame.sort_values() raising an AttributeError when sorting on a key that casts column to categorical dtype (GH 36383)
• Bug in DataFrame.stack() raising a ValueError when stacking MultiIndex columns based on position when the levels had duplicate names (GH 36353)
• Bug in Series.astype() showing too much precision when casting from np.float32 to string dtype (GH 36451)
• Bug in Series.isin() and DataFrame.isin() when using NaN and a row length above 1,000,000 (GH 22205)
• Bug in cut() raising a ValueError when passed a Series of labels with ordered=False (GH 36603)

Other#

• Reverted enhancement added in pandas-1.1.0 where timedelta_range() infers a frequency when passed start, stop, and periods (GH 32377)

Contributors#

A total of 16 people contributed patches to this release. People with a “+” by their names contributed a patch for the first time.

• Asish Mahapatra
• Dan Moore +
• Daniel Saxton
• Fangchen Li
• Hans
• Irv Lustig
• Joris Van den Bossche
• Kaiqi Dong
• MeeseeksMachine
• Number42 +
• Pandas Development Team
• Richard Shadrach
• Simon Hawkins
• jbrockmendel
• nrebena
• patrick