Releases: alteryx/evalml
Releases · alteryx/evalml
v0.83.0
v0.83.0 Feb. 2, 2024
Enhancements
- Added support for additional estimators for multiseries datasets #4385
Fixes
- Fixed bug in
_downcast_nullable_y
causing woodwork initialization issues #4369 - Fixed multiseries prediction interval labels #4377
Changes
- Pinned scipy version to under 1.12.0 #4380
Documentation Changes
Testing Changes
Breaking Changes
v0.82.0
v0.82.0 Nov. 3, 2023
Enhancements
- Changed target name/series ID divider and added ability to return series ID column with predictions #4357
Fixes
Changes
- Pinned networkx version below 3.2 for Python version compatibility #4351
Documentation Changes
- Added multiseries time series section to user guide in documentation #4355
- Updated release guide to include an FAQ section about fixing github actions #4346
Testing Changes
Breaking Changes
v0.81.1
v0.81.0
Enhancements
- Extended STLDecomposer to support multiseries #4253
- Extended TimeSeriesImputer to support multiseries #4291
- Added datacheck to check for mismatched series length in multiseries #4296
- Added STLDecomposer to multiseries pipelines #4299
- Extended DateTimeFormatCheck data check to support multiseries #4300
- Extended TimeSeriesRegularizer to support multiseries #4303
Fixes
- Fixed forecast period generation function for multiseries #4320
- Fixed bug in
STLDecomposer.inverse_transform
causing incorrect seasonality projections #4328
Changes
- Updated
split_data
to callsplit_multiseries_data
when passed stacked multiseries data #4312 - Pinned pandas version under 2.1.0 #4315
- Increased minimum numpy version #4321
Documentation Changes
- Removed LightGBM's excessive amount of warnings #4308
Testing Changes
- Removed old performance testing workflow #4318
v0.80.0
Enhancements
- Added support for prediction intervals for VARMAX regressor #4267
- Integrated multiseries time series into AutoMLSearch #4270
Fixes
- Fixed error when stacking data with no exogenous variables #4275
Changes
- Updated ARIMARegressor to be compatible with sktime v0.22.0 and beyond #4283
- Updated graph_prediction_vs_actual_over_time() to be compatible with multiseries time series #4284
- Updated excluded_model_families to take in a list of both str and ModelFamily data types #4287
- Unpinned ipywidgets #4288
Documentation Changes
- Removed erroneous warnings from Data Checks User Guide page and removed tqdm warning in all notebooks #4274
v0.79.0
Enhancements
- Updated regression metrics to handle multioutput dataframes as well as single output series #4233
- Added baseline regressor for multiseries time series problems #4246
- Added stacking and unstacking utility functions to work with multiseries data #4250
- Added multiseries regression pipeline class #4256
- Added multiseries VARMAX regressor #4238
Fixes
- Added support for pandas 2 #4216
- Fixed bug where time series pipelines would fail due to MASE needing
y_train
when scoring #4258 - Update s3 bucket for docs image #4260
- Fix deps checker including any package with post in the name #4268
Changes
v0.78.0
Enhancements
- Add run_feature_selection to AutoMLSearch and Default Algorithm #4210
- Added SMAPE to the standard metrics for time series problems #4220
Fixes
- IDColumnsDataCheck now works with Unknown data type #4203
Changes
- Upgraded minimum SHAP version to 0.42.0 and unpinned numpy version #4228
Documentation Changes
- Updated API reference #4213
v0.77.0
v0.77.0 June. 07, 2023
Enhancements
- Added
check_distribution
function for determining if the predicted distribution matches the true one :pr:4184
- Added
get_recommendation_score_breakdown
function for insight on the recommendation score :pr:4188
- Added excluded_model_families parameter to AutoMLSearch() :pr:
4196
- Added option to exclude time index in
IDColumnsDataCheck
:pr:4194
Fixes
- Fixed small errors in
ARIMARegressor
implementation :pr:4186
- Fixed
get_forecast_period
to properly handlegap
parameter :pr:4200
Testing Changes
- Run looking glass performance tests on merge via Airflow :pr:
4198
v0.76.0
v0.75.0
v0.75.0 May. 2, 2023
Fixes
- Fixed bug where resetting the holdout data indices would cause time series
predict_in_sample
to be wrong #4161