Skip to content

Releases: alteryx/evalml

v0.57.0 Sept. 6, 2022

06 Sep 19:17
d83053f
Compare
Choose a tag to compare

v0.57.0 Sept. 6, 2022

  • Enhancements
    • Added KNNImputer class and created new knn parameter for Imputer :pr:3662
  • Fixes
    • IDColumnsDataCheck now only returns an action code to set the first column as the primary key if it contains unique values :pr:3639
    • IDColumnsDataCheck now can handle primary key columns containing "integer" values that are of the double type :pr:3683
    • Added support for BooleanNullable columns in EvalML pipelines and imputer :pr:3678
    • Updated StandardScaler to only apply to numeric columns :pr:3686
  • Changes
    • Unpinned sktime to allow for version 0.13.2 :pr:3685
    • Pinned pmdarima < 2.0.0 :pr:3679

v0.56.1

19 Aug 16:47
9e2d3d0
Compare
Choose a tag to compare

v0.56.1 Aug. 19, 2022

Fixes

  • IDColumnsDataCheck now only returns an action code to set the first column as the primary key if it contains unique values #3639
  • Reverted the make_pipeline changes that conditionally included the imputers #3672

v0.56.0

16 Aug 05:17
a5f32a0
Compare
Choose a tag to compare

v0.56.0 Aug. 16, 2022

Enhancements

  • Add CI testing environment in Mac for install workflow #3646
  • Updated make_pipeline to only include the Imputer in pipelines if NaNs exist in the data #3657
  • Updated to run with Woodwork >= 0.17.2 #3626
  • Add exclude_featurizers parameter to AutoMLSearch to specify featurizers that should be excluded from all pipelines #3631
  • Add fit_transform method to pipelines and component graphs #3640
  • Changed default value of data splitting for time series problem holdout set evaluation #3650

Fixes

  • Reverted the Woodwork 0.17.x compatibility work due to performance regression #3664

Changes

  • Disable holdout set in AutoML search by default #3659
  • Pinned sktime at >=0.7.0,<0.13.1 due to slowdowns with time series modeling #3658
  • Added additional testing support for Python 3.10 #3609

Documentation Changes

  • Updated broken link checker to exclude stackoverflow domain #3633
  • Add instructions to add new users to evalml-core-feedstock #3636

v0.55.0

28 Jul 15:39
8ad38ea
Compare
Choose a tag to compare

v0.55.0 Jul. 28, 2022

Enhancements

  • Increased the amount of logical type information passed to Woodwork when calling ww.init() in transformers #3604
  • Added ability to log how long each batch and pipeline take in automl.search() #3577
  • Added the option to set the sp parameter for ARIMA models #3597
  • Updated the CV split size of time series problems to match forecast horizon for improved performance #3616

Fixes

  • Fixed iterative graphs not appearing in documentation #3592
  • Updated the load_diabetes() method to account for scikit-learn 1.1.1 changes to the dataset #3591
  • Capped woodwork version at < 0.17.0 #3612
  • Bump minimum scikit-optimize version to 0.9.0 `#3614

Changes

  • Add pre-commit hooks for linting #3608
  • Implemented a lower threshold and window size for the TimeSeriesRegularizer and DatetimeFormatDataCheck #3627

Documentation Changes

Testing Changes

  • Pinned GraphViz version for Windows CI Test #3596
  • Removed pytest.mark.skip_if_39 pytest marker #3602 #3607
  • Refactored test cases that iterate over all components to use pytest.mark.parametrise and changed the corresponding if...continue blocks to pytest.mark.xfail #3622

v0.54.0

27 Jun 17:12
d259cb9
Compare
Choose a tag to compare

v0.54.0 Jun. 27, 2022

Fixes

  • Updated the Imputer and SimpleImputer to work with scikit-learn 1.1.1. #3525
  • Bumped the minimum versions of scikit-learn to 1.1.1 and imbalanced-learn to 0.9.1. #3525
  • Added a clearer error message when describe is called on an un-instantiated ComponentGraph #3569
  • Added a clearer error message when time series' predict is called with its X_train or y_train parameter set as None #3579

Changes

  • Don't pass time_index as kwargs to sktime ARIMA implementation for compatibility with latest version #3564
  • Remove incompatible nlp-primitives version 2.6.0 from accepted dependency versions #3572, #3574
  • Updated evalml authors #3581

Documentation Changes

  • Fix typo in long_description field in setup.cfg #3553
  • Update install page to remove Python 3.7 mention #3567

v0.53.1

09 Jun 20:44
67df45a
Compare
Choose a tag to compare

v0.53.1 Jun. 9, 2022

Enhancements

Fixes

Changes

  • Set the development status to 4 - Beta in setup.cfg #3550

Documentation Changes

Testing Changes

v0.53.0

09 Jun 18:58
882e7cc
Compare
Choose a tag to compare

v0.53.0 Jun. 9, 2022

Enhancements

  • Pass n_jobs to default algorithm #3548

Fixes

  • Fixed github workflows for featuretools and woodwork to test their main branch against evalml. #3517
  • Supress warnings in TargetEncoder raised by a coming change to default parameters #3540
  • Fixed bug where schema was not being preserved in column renaming for XGBoost and LightGBM models #3496

Changes

  • Transitioned to use pyproject.toml and setup.cfg away from setup.py #3494, #3536

Documentation Changes

  • Updated the Time Series User Guide page to include known-in-advance features and fix typos #3521
  • Add slack and stackoverflow icon to footer #3528
  • Add install instructions for M1 Mac #3543

Testing Changes

  • Rename yml to yaml for GitHub Actions #3522
  • Remove noncore_dependency pytest marker #3541
  • Changed test_smotenc_category_features to use valid postal code values in response to new woodwork type validation #3544

v0.52.0

12 May 20:11
40e6325
Compare
Choose a tag to compare

v0.52.0 May. 12, 2022

Changes

  • Added github workflows for featuretools and woodwork to test their main branch against evalml. #3504
  • Added pmdarima to conda recipe. #3505
  • Added a threshold for NullDataCheck before a warning is issued for null values #3507
  • Changed NoVarianceDataCheck to only output warnings #3506
  • Updated roc_curve() and conf_matrix() to work with IntegerNullable and BooleanNullable types. #3465
  • Changed ComponentGraph._transform_features to raise a PipelineError instead of a ValueError. This is not a breaking change because PipelineError is a subclass of ValueError. #3497

Documentation Changes

  • Updated to install prophet extras in Read the Docs. #3509

Testing Changes

  • Moved vowpal wabbit in test recipe to evalml package from evalml-core #3502

v0.51.0

29 Apr 15:34
882a9e4
Compare
Choose a tag to compare

v0.51.0 Apr. 29, 2022

Enhancements

  • Updated make_pipeline_from_data_check_output to work with time series problems. #3454

Fixes

  • Changed PipelineBase.graph_json() to return a python dictionary and renamed as graph_dict()#3463

Changes

  • Added vowpalwabbit to local recipe and remove is_using_conda pytest skip markers from relevant tests #3481

Documentation Changes

  • Fixed broken link in contributing guide #3464
  • Improved development instructions #3468
  • Added the TimeSeriesRegularizer and TimeSeriesImputer to the timeseries section of the User Guide #3473
  • Updated OSS slack link #3487

Testing Changes

  • Updated unit tests to support woodwork 0.16.2 #3482
  • Fix some unit tests after vowpal wabbit got added to conda recipe #3486

Breaking Changes

  • Renamed PipelineBase.graph_json() to PipelineBase.graph_dict() #3463
  • Minimum supported woodwork version is now 0.16.2 #3482

v0.50.0

13 Apr 15:39
fa81afb
Compare
Choose a tag to compare

v0.50.0 Apr. 13, 2022

Enhancements

  • Added TimeSeriesImputer component #3374
  • Replaced pipeline_parameters and custom_hyperparameters with search_parameters in AutoMLSearch #3373, #3427
  • Added TimeSeriesRegularizer to smooth uninferrable date ranges for time series problems #3376
  • Enabled ensembling as a parameter for DefaultAlgorithm #3435, #3444

Fixes

  • Fix DefaultAlgorithm not handling Email and URL features #3419
  • Added test to ensure LabelEncoder parameters preserved during AutoMLSearch #3326

Changes

  • Updated DateTimeFormatDataCheck to use woodwork's infer_frequency function #3425
  • Renamed graphs.py to visualizations.py #3439

Documentation Changes

  • Updated the model understanding section of the user guide to include missing functions #3446
  • Rearranged the user guide model understanding page for easier navigation #3457

Breaking Changes

  • Renamed graphs.py to visualizations.py #3439
  • Replaced pipeline_parameters and custom_hyperparameters with search_parameters in AutoMLSearch #3373