v0.22.0

v0.22.0 March 13, 2023

• Enhancements
  • Improved inference for numeric logical types to handle incoming object dtype data (#1645)
  • Updated datetime format inference to handle years represented by 2 digits (#1632)
  • Updated dependence_dict to handle boolean columns (#1652)
• Changes
  • Pin for jupyter-client to 7.4.9 for documentation (#1624)
  • Remove jupyter-client documentation requirement (#1627)
  • Separate Makefile command for core requirements, test requirements and dev requirements (#1658)
• Testing Changes
  • Add ruff for linting and replace isort/flake8 (#1614)
  • Specify black and ruff config arguments (#1620)
  • Add codecov token for unit tests workflow (#1630)
  • Add GitHub Actions cache to speed up workflows (#1631)
  • Add pull request check for linked issues to CI workflow (#1633 #1636)
  • Run lint fix on latest dependency update pull requests (#1640 #1641)

Thanks to the following people for contributing to this release:
@bchen1116, @gsheni, @ParthivNaresh

v0.21.2

v0.21.2 January 11, 2022

• Changes
  • Bump scipy and scikit-learn min versions for compatibility with numpy 1.24.0 (#1606)
  • Add is_natural_language method to ColumnSchema object (#1610)
  • Changed the transform function for the Boolean logical type to improve runtime (#1612)

Thanks to the following people for contributing to this release:
@ParthivNaresh, @sbadithe, @thehomebrewnerd

v0.21.1

v0.21.1 December 16, 2022

• Fixes
  • Fix importlib DeprecationWarning in inference_functions.py (#1584)
  • Schema now maintains column order after renaming a column (#1594)
  • Fixed logic to not set config during boolean transform (#1601)
• Changes
  • Rename backup_dtype to pyspark_dtype (#1593)
  • Removed inference for ["0", "1"], ["0.0", "1.0"], and [0, 1] as Boolean logical types, but maintained forced inference of such values (#1600)

Thanks to the following people for contributing to this release:
@bchen1116, @sbadithe

v0.21.0

v0.21.0 December 1, 2022

• Enhancements
  • Improved Boolean and BooleanNullable inference to detect common string representations of boolean values (#1549)
  • Added the get_outliers and medcouple_dict functions to WoodworkColumnAccessor so that the medcouple statistic can be used for outlier detection (#1547)
• Fixes
  • Resolved FutureWarning in _get_box_plot_info_for_column (#1563)
  • Fixed error message in validate method in logical_types.py (#1565)
  • Fixed IntegerNullable inference by checking values are within valid Int64 bounds (#1572)
  • Update demo dataset links to point to new endpoint (#1570)
  • Fix DivisionByZero error in type_system.py (#1571)
  • Fix Categorical dtype inference for PostalCode logical type (#1574)
  • Fixed issue where forcing a Boolean logical type on a column of 0.0s and 1.0s caused incorrect transformation (#1576)
• Changes
  • Unpin dask dependency (#1561)
  • Changed the sampling strategy for type inference from head to random (#1566)
• Documentation Changes
  • Updated documentation to include the get_outliers and medcouple_dict (#1547)
• Testing Changes
  • Run looking glass performance tests on merge (#1567)

Breaking Changes

• (#1549) will automatically infer more values as Boolean or BooleanNullable, including, but not limited to, [0, 1], ['yes', 'no'], and ["True", "False"].

Thanks to the following people for contributing to this release:
@bchen1116, @gsheni, @ParthivNaresh, @sbadithe, @simha104

v0.20.0

v0.20.0 October 31, 2022

• Enhancements
  • Replace use of deprecated append method for dataframes and series with concat method (#1533)
• Fixes

  • Fixed bug relating to dependence calculations to ensure columns exist in dataframe (#1534)

  • Small typo fix in select docstring (#1544)

  • Fix TypeValidationError message (#1557)

  • Set dask version below 2022.10.1 (#1558)

    Thanks to the following people for contributing to this release:
    @bchen1116, @sbadithe

v0.19.0

v0.19.0 September 27, 2022

• Enhancements
  • Added Spearman Correlation to options for dependence calculations (#1523)
  • Added ignore_zeros as an argument for box_plot_dict to allow for calculations of outliers without 0 values (#1524)
  • Added target_col argument to dependence and dependence_dict to calculate correlations between features and target_col (#1531)
• Fixes
  • Fix datetime pivot point to be set at current year + 10 rather than the default for two-digit years when datetime_format provided (#1512)
• Changes
  • Added ignore_columns as an argument when initializing a dataframe (#1504)
  • Remove dask[dataframe] version restriction (#1527)
• Testing Changes
  • Add kickoff for create conda forge pull request from release (#1515)

Thanks to the following people for contributing to this release:
@bchen1116, @gsheni, @ParthivNaresh, @thehomebrewnerd

v0.18.0

v0.18.0 August 31, 2022

• Enhancements
  • Updated dependence_dict and mutual_information to drop to drop Categorical columns with a large number of unique values during mutual information calculation, non-dask only. (#1501)
• Fixes
  • Fix applying LatLong.transform to empty dask data (#1507)
• Changes
  • Transition from setup.cfg to pyproject.toml (#1506`,#1508)
  • Added a check to see if a series dtype has changed prior to using _replace_nans (#1502)
• Testing Changes
  • Update development requirements and use latest for documentation (#1499)

Thanks to the following people for contributing to this release:
@bchen1116, @gsheni, @jeff-hernandez, @ParthivNaresh, @rwedge

v0.17.2

v0.17.2 August 5, 2022

• Fixes
  • Updated concat_columns to work with dataframes with mismatched indices or different shapes (#485)
• Documentation Changes
  • Add instructions to add new users to woodwork feedstock (#483)
• Testing Changes
  • Add create feedstock PR workflow (#1489)

Thanks to the following people for contributing to this release:
@chukarsten , @Cmancuso, @gsheni

v0.17.1

v0.17.1 July 29, 2022

* Testing Changes
   * Allow for manual kickoff for minimum dependency checker (#1476)

Thanks to the following people for contributing to this release:
@bchen1116, @gsheni

v0.17.0

v0.17.0 July 14, 2022

warning: This release of Woodwork will not support Python 3.7

• Enhancements
  • Added ability to null invalid values for Double logical type #1449
  • Added ability to null invalid values for BooleanNullable logical type #1455
  • Added ability to null invalid values for IntegerNullable logical type #1456
  • Added ability to null invalid values for EmailAddress logical type #1457
  • Added ability to null invalid values for URL logical type #1459
  • Added ability to null invalid values for PhoneNumber logical type #1460
  • Added ability to null invalid values for AgeFractional and AgeNullable logical types #1462
  • Added ability to null invalid values for LatLong logical type #1465
  • Added ability to null invalid values for PostalCode logical type (US only) #1467
  • Added smarter inference for IntegerNullable and BooleanNullable types #1458
• Fixes
  • Fixed inference of all null string values as Unknown instead of Datetime #1458
• Changes
  • Set the minimum acceptable version of pandas to 1.4.0 for woodwork and 1.4.3 for spark add-on #1461
  • Dropped support for Python 3.7 #1461
  • Add pre-commit hooks for linting #1470

Thanks to the following people for contributing to this release:
@gsheni , @jeff-hernandez, @ParthivNaresh