Skip to content

ENH: Add prod to masked_reductions #33442

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 16 commits into from
Apr 12, 2020
Merged

ENH: Add prod to masked_reductions #33442

Show file tree
Hide file tree
Changes from 8 commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
3e1f06c
REF: Add prod to masked_reductions
dsaxton Apr 8, 2020
a7eb301
Update
dsaxton Apr 9, 2020
48c1432
Change test
dsaxton Apr 10, 2020
c73d7b9
Keep
dsaxton Apr 10, 2020
a429ced
Merge remote-tracking branch 'upstream/master' into masked-prod
dsaxton Apr 10, 2020
b0c95fc
Update
dsaxton Apr 10, 2020
c3e1763
Move functions
dsaxton Apr 10, 2020
6b66756
Lint
dsaxton Apr 10, 2020
58f7bd0
maybe_cast_result_dtype
dsaxton Apr 10, 2020
a2574df
Lint
dsaxton Apr 10, 2020
474506b
Merge remote-tracking branch 'upstream/master' into masked-prod
dsaxton Apr 10, 2020
57551b8
Revert and change test
dsaxton Apr 11, 2020
1d25569
Lint
dsaxton Apr 11, 2020
de5954a
Merge remote-tracking branch 'upstream/master' into masked-prod
dsaxton Apr 11, 2020
2701fff
Remove
dsaxton Apr 11, 2020
8321945
Lint
dsaxton Apr 11, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion doc/source/whatsnew/v1.1.0.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -358,7 +358,7 @@ Performance improvements
sparse values from ``scipy.sparse`` matrices using the
:meth:`DataFrame.sparse.from_spmatrix` constructor (:issue:`32821`,
:issue:`32825`, :issue:`32826`, :issue:`32856`, :issue:`32858`).
- Performance improvement in reductions (sum, min, max) for nullable (integer and boolean) dtypes (:issue:`30982`, :issue:`33261`).
- Performance improvement in reductions (sum, prod, min, max) for nullable (integer and boolean) dtypes (:issue:`30982`, :issue:`33261`, :issue:`33442`).


.. ---------------------------------------------------------------------------
Expand Down
48 changes: 31 additions & 17 deletions pandas/core/array_algos/masked_reductions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,8 @@
for missing values.
"""

from typing import Callable

import numpy as np

from pandas._libs import missing as libmissing
Expand All @@ -11,14 +13,19 @@
from pandas.core.nanops import check_below_min_count


def sum(
values: np.ndarray, mask: np.ndarray, skipna: bool = True, min_count: int = 0,
def _sumprod(
func: Callable,
values: np.ndarray,
mask: np.ndarray,
skipna: bool = True,
min_count: int = 0,
):
"""
Sum for 1D masked array.
Sum or product for 1D masked array.

Parameters
----------
func : np.sum or np.prod
values : np.ndarray
Numpy array with the values (can be of any dtype that support the
operation).
Expand All @@ -31,23 +38,33 @@ def sum(
``min_count`` non-NA values are present the result will be NA.
"""
if not skipna:
if mask.any():
if mask.any() or check_below_min_count(values.shape, None, min_count):
return libmissing.NA
else:
if check_below_min_count(values.shape, None, min_count):
return libmissing.NA
return np.sum(values)
return func(values)
else:
if check_below_min_count(values.shape, mask, min_count):
return libmissing.NA

if _np_version_under1p17:
return np.sum(values[~mask])
return func(values[~mask])
else:
return np.sum(values, where=~mask)
return func(values, where=~mask)


def sum(values: np.ndarray, mask: np.ndarray, skipna: bool = True, min_count: int = 0):
return _sumprod(
np.sum, values=values, mask=mask, skipna=skipna, min_count=min_count
)


def _minmax(func, values: np.ndarray, mask: np.ndarray, skipna: bool = True):
def prod(values: np.ndarray, mask: np.ndarray, skipna: bool = True, min_count: int = 0):
return _sumprod(
np.prod, values=values, mask=mask, skipna=skipna, min_count=min_count
)


def _minmax(func: Callable, values: np.ndarray, mask: np.ndarray, skipna: bool = True):
"""
Reduction for 1D masked array.

Expand All @@ -63,18 +80,15 @@ def _minmax(func, values: np.ndarray, mask: np.ndarray, skipna: bool = True):
Whether to skip NA.
"""
if not skipna:
if mask.any():
if mask.any() or not values.size:
# min/max with empty array raise in numpy, pandas returns NA
return libmissing.NA
else:
if values.size:
return func(values)
else:
# min/max with empty array raise in numpy, pandas returns NA
return libmissing.NA
return func(values)
else:
subset = values[~mask]
if subset.size:
return func(values[~mask])
return func(subset)
else:
# min/max with empty array raise in numpy, pandas returns NA
return libmissing.NA
Expand Down
19 changes: 11 additions & 8 deletions pandas/core/arrays/boolean.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -696,9 +696,18 @@ def _reduce(self, name: str, skipna: bool = True, **kwargs):
data = self._data
mask = self._mask

if name in {"sum", "min", "max"}:
if name in {"sum", "prod", "min", "max"}:
op = getattr(masked_reductions, name)
return op(data, mask, skipna=skipna, **kwargs)
result = op(data, mask, skipna=skipna, **kwargs)

Copy link
Contributor

@jreback jreback Apr 10, 2020 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can you try using maybe_cast_result_dtype here (you will have to add the prod operation in side that)

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As far as I understand, that should not be needed. Numpy should give the desired result for booleans.

# if we have numeric op that would result in an int,
# coerce to int if possible
if name == "prod" and notna(result):
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why did you need to add this back?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was getting a failure on some builds for \tests\arrays\boolean\test_reduction.py where we're checking that the product has int dtype so it might be needed:

        elif op == "prod":
>           assert isinstance(getattr(s, op)(), np.int64)
E           AssertionError: assert False

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, then maybe the test needs to be edited. In any case, we should investigate why it is failing / what we should be expecting.

From a quick test, it seems that numpy returns int64:

In [6]: np.array([], dtype="bool").prod()                                                                                                                                                                          
Out[6]: 1

In [7]: type(_)                                                                                                                                                                                                    
Out[7]: numpy.int64

In [8]: np.array([True, False], dtype="bool").prod()                                                                                                                                                               
Out[8]: 0

In [9]: type(_)                                                                                                                                                                                                    
Out[9]: numpy.int64

So I think we should follow that behaviour here

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

From what I can tell it was something that was only affecting Windows builds: https://dev.azure.com/pandas-dev/pandas/_build/results?buildId=32986&view=logs&jobId=077026cf-93c0-54aa-45e0-9996ba75f6f7&j=077026cf-93c0-54aa-45e0-9996ba75f6f7&t=e95cf409-86ae-5b4d-6c5f-79395ef75e8f

int_result = np.int64(result)
if int_result == result:
result = int_result

return result

# coerce to a nan-aware float if needed
if self._hasna:
Expand All @@ -710,12 +719,6 @@ def _reduce(self, name: str, skipna: bool = True, **kwargs):
if np.isnan(result):
return libmissing.NA

# if we have numeric op that would result in an int, coerce to int if possible
if name == "prod" and notna(result):
int_result = np.int64(result)
if int_result == result:
result = int_result

return result

def _maybe_mask_result(self, result, mask, other, op_name: str):
Expand Down
9 changes: 1 addition & 8 deletions pandas/core/arrays/integer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,6 @@

from pandas.core import nanops, ops
from pandas.core.array_algos import masked_reductions
import pandas.core.common as com
from pandas.core.indexers import check_array_indexer
from pandas.core.ops import invalid_comparison
from pandas.core.ops.common import unpack_zerodim_and_defer
Expand Down Expand Up @@ -562,7 +561,7 @@ def _reduce(self, name: str, skipna: bool = True, **kwargs):
data = self._data
mask = self._mask

if name in {"sum", "min", "max"}:
if name in {"sum", "prod", "min", "max"}:
op = getattr(masked_reductions, name)
return op(data, mask, skipna=skipna, **kwargs)

Expand All @@ -581,12 +580,6 @@ def _reduce(self, name: str, skipna: bool = True, **kwargs):
if name in ["any", "all"]:
pass

# if we have a preservable numeric op,
# provide coercion back to an integer type if possible
elif name == "prod":
# GH#31409 more performant than casting-then-checking
result = com.cast_scalar_indexer(result)

return result

def _maybe_mask_result(self, result, mask, other, op_name: str):
Expand Down
2 changes: 1 addition & 1 deletion pandas/tests/arrays/integer/test_dtypes.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ def test_preserve_dtypes(op):

# op
result = getattr(df.C, op)()
if op in {"sum", "min", "max"}:
if op in {"sum", "prod", "min", "max"}:
assert isinstance(result, np.int64)
else:
assert isinstance(result, int)
Expand Down
5 changes: 3 additions & 2 deletions pandas/tests/extension/test_integer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -238,9 +238,10 @@ def check_reduce(self, s, op_name, skipna):
# overwrite to ensure pd.NA is tested instead of np.nan
# https://github.com/pandas-dev/pandas/issues/30958
result = getattr(s, op_name)(skipna=skipna)
expected = getattr(s.astype("float64"), op_name)(skipna=skipna)
if np.isnan(expected):
if not skipna and s.isna().any():
expected = pd.NA
else:
expected = getattr(s.dropna().astype("int64"), op_name)(skipna=skipna)
tm.assert_almost_equal(result, expected)


Expand Down