Skip to content
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

BUG: Adding skipna as an option to groupby cumsum and cumprod #19914

Merged
merged 10 commits into from
Mar 1, 2018
Merged

BUG: Adding skipna as an option to groupby cumsum and cumprod #19914

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
90b7978
Adding skipna as an option to groupby cumsum and cumprod
shangyian Feb 26, 2018
84db344
Fixing flake8 flags
shangyian Feb 26, 2018
6182352
Adding whatsnew entry
shangyian Feb 26, 2018
2579eaa
Adding a better test
shangyian Feb 26, 2018
75d2870
Fixing pep8 issues
shangyian Feb 26, 2018
bb740fd
Making some changes based on review and altering tests
shangyian Feb 28, 2018
47fe8d6
Making test changes based on review
shangyian Mar 1, 2018
107bc0b
More changes to tests based on review
shangyian Mar 1, 2018
5aceda9
Merge branch 'master' into PR_TOOL_MERGE_PR_19914
jreback Mar 1, 2018
3072df7
doc
jreback Mar 1, 2018
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
1 change: 1 addition & 0 deletions doc/source/whatsnew/v0.23.0.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -890,6 +890,7 @@ Groupby/Resample/Rolling
- Bug in :func:`DataFrame.groupby` passing the `on=` kwarg, and subsequently using ``.apply()`` (:issue:`17813`)
- Bug in :func:`DataFrame.resample().aggregate` not raising a ``KeyError`` when aggregating a non-existent column (:issue:`16766`, :issue:`19566`)
- Fixed a performance regression for ``GroupBy.nth`` and ``GroupBy.last`` with some object columns (:issue:`19283`)
- Bug in :func:`DataFrame.groupby.cumsum` and :func:`DataFrame.groupby.cumprod` where skipna is not an option
Copy link
Contributor

Choose a reason for hiding this comment

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

These should be

:meth:`DataFrameGroupBy.cumsum`

likewise for cumprod.

And skipna should probably be in double backticks.

And could you add a :issue: for the issue you're closing.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Added those in


Sparse
^^^^^^
Expand Down
18 changes: 16 additions & 2 deletions pandas/_libs/groupby.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -139,7 +139,8 @@ def group_median_float64(ndarray[float64_t, ndim=2] out,
def group_cumprod_float64(float64_t[:, :] out,
float64_t[:, :] values,
int64_t[:] labels,
bint is_datetimelike):
bint is_datetimelike,
bint skipna=True):
"""
Only transforms on axis=0
"""
Expand All @@ -163,14 +164,21 @@ def group_cumprod_float64(float64_t[:, :] out,
if val == val:
accum[lab, j] *= val
out[i, j] = accum[lab, j]
if val != val:
Copy link
Contributor

Choose a reason for hiding this comment

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

This can just be else?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Yeah, that makes more sense. I changed that for both cumsum and cumprod.

if skipna:
out[i, j] = NaN
else:
accum[lab, j] = NaN
out[i, j] = accum[lab, j]
Copy link
Contributor

Choose a reason for hiding this comment

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

Can you break early here? If you ever hit a NaN in a group, final output will be NaN so no need to continue adding to NaN.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

That's true - I added a break there.



@cython.boundscheck(False)
@cython.wraparound(False)
def group_cumsum(numeric[:, :] out,
numeric[:, :] values,
int64_t[:] labels,
is_datetimelike):
is_datetimelike,
bint skipna=True):
"""
Only transforms on axis=0
"""
Expand All @@ -196,6 +204,12 @@ def group_cumsum(numeric[:, :] out,
if val == val:
accum[lab, j] += val
out[i, j] = accum[lab, j]
if val != val:
Copy link
Contributor

Choose a reason for hiding this comment

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

Same comments about if -> else and early breaking.

if skipna:
out[i, j] = NaN
else:
accum[lab, j] = NaN
out[i, j] = accum[lab, j]
else:
accum[lab, j] += val
out[i, j] = accum[lab, j]
Expand Down
6 changes: 4 additions & 2 deletions pandas/core/groupby.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1839,7 +1839,8 @@ def rank(self, method='average', ascending=True, na_option='keep',
@Appender(_doc_template)
def cumprod(self, axis=0, *args, **kwargs):
"""Cumulative product for each group"""
nv.validate_groupby_func('cumprod', args, kwargs, ['numeric_only'])
nv.validate_groupby_func('cumprod', args, kwargs,
['numeric_only', 'skipna'])
if axis != 0:
return self.apply(lambda x: x.cumprod(axis=axis, **kwargs))

Expand All @@ -1849,7 +1850,8 @@ def cumprod(self, axis=0, *args, **kwargs):
@Appender(_doc_template)
def cumsum(self, axis=0, *args, **kwargs):
"""Cumulative sum for each group"""
nv.validate_groupby_func('cumsum', args, kwargs, ['numeric_only'])
nv.validate_groupby_func('cumsum', args, kwargs,
['numeric_only', 'skipna'])
if axis != 0:
return self.apply(lambda x: x.cumsum(axis=axis, **kwargs))

Expand Down
20 changes: 20 additions & 0 deletions pandas/tests/groupby/test_groupby.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2521,6 +2521,26 @@ def test_groupby_cumprod(self):
expected.name = 'value'
tm.assert_series_equal(actual, expected)

# make sure that skipna works
Copy link
Contributor

Choose a reason for hiding this comment

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

Can you split this into a new test? Make sure to add a reference to the Github issue as a comment.

Copy link
Contributor

Choose a reason for hiding this comment

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

I'd find this a bit easier to read as

df = pd.DataFrame({"x": [...], "gp": ['a'] * 6 + ['b'] * 6})

instead of concat.

Copy link
Contributor

Choose a reason for hiding this comment

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

And maybe have a group with all-NA to ensure that is handled properly.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

I added a new test (test_groupby_cum_skipna) for all of the skipna-related cases. And added a groupby+cumsum/prod on all-NA values.

df = pd.concat(
[pd.DataFrame({'x': [1.0, 2.0, np.nan, np.nan, 3.0, 2.0],
'gp': 'a'}),
pd.DataFrame({'x': [2.0, 5.0, 6.0, 1.0, np.nan, 1.0],
'gp': 'b'})])
result = df.groupby('gp')['x'].cumprod(skipna=False)
expected = pd.Series([1.0, 2.0, np.nan, np.nan, np.nan, np.nan,
2.0, 10.0, 60.0, 60.0, np.nan, np.nan],
name='x', index=(0, 1, 2, 3, 4, 5,
0, 1, 2, 3, 4, 5))
tm.assert_series_equal(result, expected)

result = df.groupby('gp')['x'].cumprod()
Copy link
Contributor

Choose a reason for hiding this comment

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

Don't think you need this test. skipna=False should be adequately tested as is.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Yep, removed it.

expected = pd.Series([1.0, 2.0, np.nan, np.nan, 6.0, 12.0,
2.0, 10.0, 60.0, 60.0, np.nan, 60.0],
name='x', index=(0, 1, 2, 3, 4, 5,
0, 1, 2, 3, 4, 5))
tm.assert_series_equal(result, expected)

Copy link
Contributor

Choose a reason for hiding this comment

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

Could you add a test for cumsum with skipna=False`?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

That should be covered by test_groupby_cum_skipna now.

def test_ops_general(self):
ops = [('mean', np.mean),
('median', np.median),
Expand Down