PERF: join/merge on subset of MultiIndex #48611
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
phofl
reviewed
Sep 17, 2022
pandas/core/reshape/merge.py
Outdated
| if dropped_level_name in left.names: | ||
| if lindexer is None: |
There was a problem hiding this comment.
Why are you pulling this inside the loop? This is harder to read than before
There was a problem hiding this comment.
Agreed. We can actually avoid calling take_nd when the indexers are None as that means "take everything". I made another commit which further improves times and should be clearer.
before after ratio
[a712c501] [ad3f42b5]
<main> <multiindex-join-subset>
- 43.8±1ms 10.5±1ms 0.24 join_merge.JoinMultiindexSubset.time_join_multiindex_subset
mroeschke
reviewed
Sep 19, 2022
mroeschke
added
Performance
phofl
approved these changes
Sep 20, 2022
|
thx @lukemanley |
phofl
pushed a commit
to phofl/pandas
that referenced
this pull request
Sep 22, 2022
mroeschke
added a commit
that referenced
this pull request
Sep 26, 2022
…8662) * BUG: Series.getitem not falling back to positional for bool index * Update pandas/tests/series/indexing/test_getitem.py Co-authored-by: Matthew Roeschke <[email protected]> * Fix build warning for use of `strdup` in ultrajson (#48369) * WEB: Update versions json to fix version switcher in the docs (#48655) * PERF: join/merge on subset of MultiIndex (#48611) * DOC: Update documentation for date_range(), bdate_range(), and interval_range() to include timedelta as a possible data type for the freq parameter (#48631) * Update documentation for date_range(), bdate_range(), and interval_range() to include timedelta as a possible data type for the freq parameter * Add test case for date_range construction using datetime.timedelta * TYP: tighten Axis (#48612) * TYP: tighten Axis * allow 'rows' * BUG: Fix metadata propagation in df.corr and df.cov, GH28283 (#48616) * Add finalize to df.corr and df.cov * Clean * TST: add test case for PeriodIndex in HDFStore(GH7796) (#48618) * TST: add test case for PeriodIndex in HDFStore * TST: add test case for PeriodIndex in HDFStore * use pytest.mark.parameterize instead * Add OpenSSF Scorecards GitHub Action (#48570) * Create scorecards.yml * Update scorecards.yml * Add OpenSSF Scorecards badge to README.md * Trim whitespace in scorecards.yml * Skip scorecards.yml on forks * Fix whitespace * Pin scorecards.yml dependencies to major versions * ENH: move an exception and add a prehook to check for exception place… (#48088) * ENH: move an exception and add a prehook to check for exception placement * ENH: fix import * ENH: revert moving error * ENH: add docstring and fix import for test * ENH: re-design approach based on feedback * ENH: update whatsnew rst * ENH: apply feedback changes * ENH: refactor to remove exception_warning_list and ignore _version.py * ENH: remove NotThisMethod from tests and all * REGR: TextIOWrapper raising an error in read_csv (#48651) * REGR: TextIOWrapper raising an error in read_csv * pyupgrade * do not try to seek on unseekable buffers * unseekable buffer might also have read ahead * safer alternative: do not mess with internal/private(?) buffer of TextIOWrapper (effectively applies the shortcut only to files pandas opens) * Fix scorecard.yml workflow (#48668) * Set scorecard-action to v2.0.3 scorecard-action does not have a major version tag. Temporarily disabling github.repository check to ensure action now works. * Enable github.repository check * BUG: DatetimeIndex ignoring explicit tz=None (#48659) * BUG: DatetimeIndex ignoring explicit tz=None * GH ref * Corrected pd.merge indicator type hint (#48677) * Corrected pd.merge indicator type hint https://pandas.pydata.org/docs/reference/api/pandas.merge.html It should be "str | bool" instead of just string * Update merge.py fixed type hint in merge.py * Update merge.py Update indicator type hint in _MergeOperation * Update merge.py Added type hint _MergeOperation init * DOC: Document default value for options.display.max_cols when not running in terminal (#48672) DOC: Document default value for options.display.max_cols display.max_cols has a default value of 20 when not running in a terminal such as Jupyter Notebook * ENH: DTA/TDA add datetimelike scalar with mismatched reso (#48669) * ENH: DTA/TDA add datetimelike scalar with mismatched reso * mypy fixup * REF: support reso in remaining tslibs helpers (#48661) * REF: support reso in remaining tslibs helpers * update setup.py * PERF: Avoid fragmentation of DataFrame in read_sas (#48603) * PERF: Avoid fragmentation of DataFrame in read_sas * Add whatsnew * Add warning * DOC: Add deprecation infos to deprecated functions (#48599) * DOC: Add deprecation infos to deprecated functions * Add sections * Fix * BLD: Build wheels using cibuildwheel (#48283) * BLD: Build wheels using cibuildwheel * update from code review Co-Authored-By: Matthew Roeschke <[email protected]> * fix 3.11 version * changes from code review * Update test_wheels.py * sync run time with pandas-wheels Co-authored-by: Matthew Roeschke <[email protected]> * REGR: Performance decrease in factorize (#48620) * TYP: type all arguments with str default values (#48508) * TYP: type all arguments with str default values * na_rep: back to str * na(t)_rep is always a string * add float for some functions * and the same for the few float default arguments * define a few more literal constants * avoid itertools.cycle mypy error * revert mistake * TST: Catch more pyarrow PerformanceWarnings (#48699) * REGR: to_hdf raising AssertionError with boolean index (#48696) * REGR: to_hdf raising AssertionError with boolean index * Add gh ref * REGR: Regression in DataFrame.loc when setting df with all True indexer (#48711) * BUG: pivot_table raising for nullable dtype and margins (#48714) * TST: Address MPL 3.6 deprecation warnings (#48695) * TST: Address MPL 3.6 deprecation warnings * Address min build * missing () Co-authored-by: Matthew Roeschke <[email protected]> Co-authored-by: Ralf Gommers <[email protected]> Co-authored-by: Marc Garcia <[email protected]> Co-authored-by: Luke Manley <[email protected]> Co-authored-by: Siddhartha Gandhi <[email protected]> Co-authored-by: Torsten Wörtwein <[email protected]> Co-authored-by: Xiao Yuan <[email protected]> Co-authored-by: paradox-lab <[email protected]> Co-authored-by: Pedro Nacht <[email protected]> Co-authored-by: dataxerik <[email protected]> Co-authored-by: jbrockmendel <[email protected]> Co-authored-by: Pablo <[email protected]> Co-authored-by: tmoschou <[email protected]> Co-authored-by: Thomas Li <[email protected]> Co-authored-by: Richard Shadrach <[email protected]>
noatamir
pushed a commit
to noatamir/pandas
that referenced
this pull request
Nov 9, 2022
noatamir
pushed a commit
to noatamir/pandas
that referenced
this pull request
Nov 9, 2022
…ndas-dev#48662) * BUG: Series.getitem not falling back to positional for bool index * Update pandas/tests/series/indexing/test_getitem.py Co-authored-by: Matthew Roeschke <[email protected]> * Fix build warning for use of `strdup` in ultrajson (pandas-dev#48369) * WEB: Update versions json to fix version switcher in the docs (pandas-dev#48655) * PERF: join/merge on subset of MultiIndex (pandas-dev#48611) * DOC: Update documentation for date_range(), bdate_range(), and interval_range() to include timedelta as a possible data type for the freq parameter (pandas-dev#48631) * Update documentation for date_range(), bdate_range(), and interval_range() to include timedelta as a possible data type for the freq parameter * Add test case for date_range construction using datetime.timedelta * TYP: tighten Axis (pandas-dev#48612) * TYP: tighten Axis * allow 'rows' * BUG: Fix metadata propagation in df.corr and df.cov, GH28283 (pandas-dev#48616) * Add finalize to df.corr and df.cov * Clean * TST: add test case for PeriodIndex in HDFStore(GH7796) (pandas-dev#48618) * TST: add test case for PeriodIndex in HDFStore * TST: add test case for PeriodIndex in HDFStore * use pytest.mark.parameterize instead * Add OpenSSF Scorecards GitHub Action (pandas-dev#48570) * Create scorecards.yml * Update scorecards.yml * Add OpenSSF Scorecards badge to README.md * Trim whitespace in scorecards.yml * Skip scorecards.yml on forks * Fix whitespace * Pin scorecards.yml dependencies to major versions * ENH: move an exception and add a prehook to check for exception place… (pandas-dev#48088) * ENH: move an exception and add a prehook to check for exception placement * ENH: fix import * ENH: revert moving error * ENH: add docstring and fix import for test * ENH: re-design approach based on feedback * ENH: update whatsnew rst * ENH: apply feedback changes * ENH: refactor to remove exception_warning_list and ignore _version.py * ENH: remove NotThisMethod from tests and all * REGR: TextIOWrapper raising an error in read_csv (pandas-dev#48651) * REGR: TextIOWrapper raising an error in read_csv * pyupgrade * do not try to seek on unseekable buffers * unseekable buffer might also have read ahead * safer alternative: do not mess with internal/private(?) buffer of TextIOWrapper (effectively applies the shortcut only to files pandas opens) * Fix scorecard.yml workflow (pandas-dev#48668) * Set scorecard-action to v2.0.3 scorecard-action does not have a major version tag. Temporarily disabling github.repository check to ensure action now works. * Enable github.repository check * BUG: DatetimeIndex ignoring explicit tz=None (pandas-dev#48659) * BUG: DatetimeIndex ignoring explicit tz=None * GH ref * Corrected pd.merge indicator type hint (pandas-dev#48677) * Corrected pd.merge indicator type hint https://pandas.pydata.org/docs/reference/api/pandas.merge.html It should be "str | bool" instead of just string * Update merge.py fixed type hint in merge.py * Update merge.py Update indicator type hint in _MergeOperation * Update merge.py Added type hint _MergeOperation init * DOC: Document default value for options.display.max_cols when not running in terminal (pandas-dev#48672) DOC: Document default value for options.display.max_cols display.max_cols has a default value of 20 when not running in a terminal such as Jupyter Notebook * ENH: DTA/TDA add datetimelike scalar with mismatched reso (pandas-dev#48669) * ENH: DTA/TDA add datetimelike scalar with mismatched reso * mypy fixup * REF: support reso in remaining tslibs helpers (pandas-dev#48661) * REF: support reso in remaining tslibs helpers * update setup.py * PERF: Avoid fragmentation of DataFrame in read_sas (pandas-dev#48603) * PERF: Avoid fragmentation of DataFrame in read_sas * Add whatsnew * Add warning * DOC: Add deprecation infos to deprecated functions (pandas-dev#48599) * DOC: Add deprecation infos to deprecated functions * Add sections * Fix * BLD: Build wheels using cibuildwheel (pandas-dev#48283) * BLD: Build wheels using cibuildwheel * update from code review Co-Authored-By: Matthew Roeschke <[email protected]> * fix 3.11 version * changes from code review * Update test_wheels.py * sync run time with pandas-wheels Co-authored-by: Matthew Roeschke <[email protected]> * REGR: Performance decrease in factorize (pandas-dev#48620) * TYP: type all arguments with str default values (pandas-dev#48508) * TYP: type all arguments with str default values * na_rep: back to str * na(t)_rep is always a string * add float for some functions * and the same for the few float default arguments * define a few more literal constants * avoid itertools.cycle mypy error * revert mistake * TST: Catch more pyarrow PerformanceWarnings (pandas-dev#48699) * REGR: to_hdf raising AssertionError with boolean index (pandas-dev#48696) * REGR: to_hdf raising AssertionError with boolean index * Add gh ref * REGR: Regression in DataFrame.loc when setting df with all True indexer (pandas-dev#48711) * BUG: pivot_table raising for nullable dtype and margins (pandas-dev#48714) * TST: Address MPL 3.6 deprecation warnings (pandas-dev#48695) * TST: Address MPL 3.6 deprecation warnings * Address min build * missing () Co-authored-by: Matthew Roeschke <[email protected]> Co-authored-by: Ralf Gommers <[email protected]> Co-authored-by: Marc Garcia <[email protected]> Co-authored-by: Luke Manley <[email protected]> Co-authored-by: Siddhartha Gandhi <[email protected]> Co-authored-by: Torsten Wörtwein <[email protected]> Co-authored-by: Xiao Yuan <[email protected]> Co-authored-by: paradox-lab <[email protected]> Co-authored-by: Pedro Nacht <[email protected]> Co-authored-by: dataxerik <[email protected]> Co-authored-by: jbrockmendel <[email protected]> Co-authored-by: Pablo <[email protected]> Co-authored-by: tmoschou <[email protected]> Co-authored-by: Thomas Li <[email protected]> Co-authored-by: Richard Shadrach <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
doc/source/whatsnew/v1.6.0.rstfile if fixing a bug or adding a new feature.Existing code passes a range to algos.take_nd. Passing an ndarray is faster.