aeon v0.5.0

See our changelog to view all changes for this release.

Following this release the deprecation policy remains suspended. Future releases may have breaking changes, so it may be wise to set an upper bound on the package version. It is likely a deprecation policy will be implemented and enforced starting v0.6.0, however.

Highlights

• New distance measures with ADTW and ShapeDTW (@chrisholder, @hadifawaz1999)
• Multiple new functions in the benchmarking module for creating figures i.e. Critical difference diagrams (@dguijo)
• New example notebooks and tidied versions of existing ones
• Lots of new bugfixes and testing to keep our estimators bug free (as much as possible at least)!

New Contributors

• @sylvaincom made their first contribution in #671
• @hedeershowk made their first contribution in #739
• @steenrotsman made their first contribution in #746
• @kgmuzungu made their first contribution in #745
• @kevinlu1248 made their first contribution in #777

All Contributors

@chrisholder, @dguijo, @hadifawaz1999, @haskarb, @hedeershowk, @kevinlu1248, @kgmuzungu, @MatthewMiddlehurst, @patrickzib, @steenrotsman, @sylvaincom, @TonyBagnall

Full Changelog: v0.4.0...v0.5.0

aeon v0.4.0

See our changelog to view all changes for this release.

Following this release the deprecation policy remains suspended. Future releases may have breaking changes, so it may be wise to set an upper bound on the package version.

Highlights

• Data downloading functions for popular classification, regression and forecasting archives are now available
• Implementations for the RSTSF interval-based and the RDST shapelet-based classifiers have been added to the classification module
• Time series adaptations for the PAM, CLARA, CLARANS clustering algorithms have been added to the clustering module
• The interval-based forests in the classification module have been reworked to use a new base class, including speed-ups and also allowing the implementation of regression versions (RISERegressor, CIFRegressor and DrCIFRegressor)
• ResNet and FCN deep learning regressors are now available
• Large portions of the website documentation have been overhauled, including the examples page and introduction notebooks for data types and data loading

All Contributors

@baraline, @chrisholder, @hadifawaz1999, @MatthewMiddlehurst, @TonyBagnall

Full Changelog: v0.3.0...v0.4.0

aeon v0.3.0

See our changelog to view all changes for this release.

Following this release the deprecation policy remains suspended. Future releases may have breaking changes, so it may be wise to set an upper bound on the package version.

Highlights

• An interface to the MrSQM algorithm has been added to the classification module.
• k-NN estimators and the Elastic Ensemble classifier now support unequal length series.
• The SAX transformation has been refactored to improve performance.
• A new collection transformer base class has been added to the transformations module for more efficient transformation using collections of time series.
• A rework of the benchmarking module has begun, starting with the introduction of functionality from kotsu

New Contributors

• @DBCerigo made their first contribution in #379
• @RafaAyGar made their first contribution in #335

All Contributors

@chrisholder, @DBCerigo, @GuiArcencio, @hadifawaz1999, @MatthewMiddlehurst, @RafaAyGar, @TonyBagnall

Full Changelog: v0.2.0...v0.3.0

aeon v0.2.0

See our changelog to view all changes for this release.

Following this release the deprecation policy remains suspended. Future releases may have breaking changes, so it may be wise to set an upper bound on the package version.

Highlights

• aeon now supports Python 3.11!
• New estimators are available in the regression package, including InceptionTime (@hadifawaz1999) and FreshPRINCE (@dguijo)
• The distances module has been reworked, and the distances available are now faster (@chrisholder)
• The RandomDilatedShapeletTransform for collections of series is now available (@baraline)
• The 'Getting Started' page on the documentation has been rewritten with clearer introductions to each module

New Contributors

• @baraline made their first contribution in #310
• @GuzalBulatova made their first contribution in #404
• @dguijo made their first contribution in #384

All Contributors

@baraline, @chrisholder, @dguijo, @GuiArcencio, @GuzalBulatova, @hadifawaz1999, @lmmentel, @MatthewMiddlehurst, @TonyBagnall

Full Changelog: v0.1.0...v0.2.0

aeon v0.1.0

See our changelog to view all changes for this release.

Following this release the deprecation policy remains suspended. Future releases may have breaking changes, so it may be wise to set an upper bound on the package version.

Highlights

• aeon is now available on PyPI!
• pandas 2 support is available for core functionality
• Deep learning approaches in the classification module have been reworked and are more configurable
• New estimators for classification in Inception Time (@hadifawaz1999) and WEASEL 2.0 (@patrickzib)
• Improved transformers for selecting channels of multivariate time series (@haskarb)

New Contributors

• @MatthewMiddlehurst made their first contribution in #10
• @aiwalter made their first contribution in #13
• @TonyBagnall made their first contribution in #12
• @patrickzib made their first contribution in #58
• @lmmentel made their first contribution in #60
• @GuiArcencio made their first contribution in #66
• @scorcism made their first contribution in #120
• @chrisholder made their first contribution in #132
• @haskarb made their first contribution in #135
• @hadifawaz1999 made their first contribution in #134

All Contributors

@aiwalter, @chrisholder, @GuiArcencio, @hadifawaz1999, @haskarb, @lmmentel, @MatthewMiddlehurst, @patrickzib, @scorcism, @TonyBagnall

Full Changelog: sktime-v0.16.0...v0.1.0

aeon v0.1.0rc0

This is an early pre-release of aeon.

This is mainly for testing purposes, a full changelog of changes from sktime v0.16.0 will be provided in a later version or the release proper.

To install aeon pre-releases from PyPI, use:

python -m pip install --pre aeon

Auto generated changelog:

What's Changed

🚀 Features

• [ENH] Dictionary Classifiers by @patrickzib in #58
• [ENH] _fit_transform method in BaseTransformer by @MatthewMiddlehurst in #91
• [ENH] Add prefer="threads" to classification Parallel usage by @MatthewMiddlehurst in #93
• [ENH] Speedup EAgglo by factor 5-10x by @patrickzib in #139
• [ENH] Adds WEASEL v2 (with dilation) by @patrickzib in #160

🐛 Bug Fixes

• [BUG] Update check_n_jobs by @MatthewMiddlehurst in #14
• [BUG] Fix Imputer bugs by @aiwalter in #24
• [BUG] Remove ComposableTimeSeriesForestRegressor by @MatthewMiddlehurst in #92
• [BUG] Fix tag in FeatureSelection and added tests by @aiwalter in #13

Other Changes

• Update FUNDING.yml by @MatthewMiddlehurst in #10
• Comment out all GitHub actions workflows. by @MatthewMiddlehurst in #3
• Remove workflow files by @MatthewMiddlehurst in #15
• Delete contrib (and update codeowners) by @TonyBagnall in #12
• [GOV] Revert CoC to remove FK changes by @aiwalter in #17
• [GOV] Removed 7 days discussion period before votings. Removed weekend rule by adding 2 days instead by @aiwalter in #41
• [GOV] CC and CoC term limitation to 2 years by @aiwalter in #22
• [FORK] Update README (IN PROGRESS) by @MatthewMiddlehurst in #43
• [GOV] Updated team page by @aiwalter in #19
• [FORK] Revert #15 "Remove workflow files" (MERGE WHEN REPO PUBLIC) by @MatthewMiddlehurst in #16
• License Update by @MatthewMiddlehurst in #56
• [GOV] Added CoCC members as voted by core developers by @aiwalter in #57
• [MNT] remove shellcheck from pre-commit checks by @lmmentel in #60
• [DOC] Docs disclaimer by @MatthewMiddlehurst in #64
• [MNT] Bump versions of pre-commit checks by @lmmentel in #62
• [GOV] Removed CC Observer role by @aiwalter in #49
• [ENH] kNN Classifier and Regressor reimplementations by @GuiArcencio in #66
• [BUG] Fix documentation build errors by @lmmentel in #73
• [FORK] Update config files by @MatthewMiddlehurst in #48
• [GOV] proposal: CC and CoC should have disjoint membership by @aiwalter in #23
• [MNT] Fix wrong mail by @aiwalter in #83
• [DOC] Classifier docs tidy up by @TonyBagnall in #52
• Update README by @MatthewMiddlehurst in #63
• [ENH] Remove ProximityForest classifier by @TonyBagnall in #86
• [Bug] ClaSP Bugfixes by @patrickzib in #95
• [MNT] convolution_based rename by @MatthewMiddlehurst in #90
• [GOV] Appointment of CC and CoC members by @aiwalter in #20
• [ENH] remove TimeSeriesSVC by @TonyBagnall in #105
• [ENH] Remove the alignment module by @TonyBagnall in #87
• [ENH] Changed grid search parallelization to use backend param from e… by @aiwalter in #82
• [MNT] Remove hcrystalball wrapper by @aiwalter in #115
• Remove duplicate function by @scorcism in #120
• [ENH] Remove fit_predict_proba from Base-Class, Use sklearn cross_val_predict instead by @patrickzib in #117
• [ENH] make single problem loaders for equal length problems return numpy arrays by @TonyBagnall in #109
• [ENH] Remove plotting by @TonyBagnall in #127
• [MNT] removed separate mlflow dependencies and testing by @aiwalter in #81
• [ENH] Early classification updates by @MatthewMiddlehurst in #88
• [DOC] Fix docstring of TabularToSeriesAdaptor by @aiwalter in #131
• [ENH] Refactored PCATransformer by @aiwalter in #26
• [ENH] Classifier tests for capability:contractable and capability:train_estimate tags by @MatthewMiddlehurst in #89
• [FORK] removed discord links by @aiwalter in #137
• [FORK] Changed all Slack links by @aiwalter in #138
• [ENH] Clean up clustering module by @chrisholder in #132
• Remove the DistanceFeatures transformer by @TonyBagnall in #143
• [ENH] [BUG] Improving Runtime of Elastic Ensemble by 20x (by fixing the closure issue) by @patrickzib in #142
• [MNT] Remove Python 3.7 support by @aiwalter in #103
• [ENH] Add tags for estimators using the taxonomy of classification by @TonyBagnall in #129
• [DOC] Rename some occurrences of sktime to aeon by @MatthewMiddlehurst in #161
• [ENH] Refactor ColumnEnsemble to ChannelEnsemble by @TonyBagnall in #162
• [ENH] Switch from pandas iteritems() to items() due to warning by @TonyBagnall in #165
• [ENH] Refactor DummyClassifier and Compose classifiers by @TonyBagnall in #163
• [MNT] Update name to aeon in config files by @MatthewMiddlehurst in #157
• [ENH] Remove old Transformer Minxin Classes by @MatthewMiddlehurst in #169
• [ENH]Update Channel Selection code by @haskarb in #135
• [MNT] Fix random test failure in FeatureSelection by @aiwalter in #180
• [ENH] Removes the dist_kerns module by @TonyBagnall in #179
• [MNT] Pandas 2.0.0 deprecation actions (part 1) by @aiwalter in #153
• [ENH] For classification: Adding InceptionTime, CNN and ResNet along with other classifiers and removing LSTM-FCN by @hadifawaz1999 in #134
• [MNT] Rename package to aeon by @MatthewMiddlehurst in #184
• [ENH] first rework of the classification notebook by @TonyBagnall in #123
• [MNT] Automated refactorings by @lmmentel in #190
• [DOC] Update landing page by @MatthewMiddlehurst in #122
• [ENH] Switch SFA Transformer to use numpy by @TonyBagnall in #196
• [ENH] Shapelet Transform improvements by @TonyBagnall in #185
• [DOC] Add website warning again. by @MatthewMiddlehurst in #205
• [MNT] Fix pytest by setting upper bound on dash by @MatthewMiddlehurst in #206
• Pre-release 0.1.0rc0 by @MatthewMiddlehurst in #207
• [MNT] Pandas 2.0.0 deprecation actions (part 2) by @aiwalter in #189
• [BUG] added missing dl tag to inceptionTime classifier by @hadifawaz1999 in #203
• [ENH] switch paa tests to use numpy by @TonyBagnall in #199

Full Changelog: sktime-v0.16.0...v0.1.0rc0

New Contributors

• @MatthewMiddlehurst made their first contribution in #10
• @aiwalter made their first contribution in #13
• @TonyBagnall made their first contribution in #12
• @patrickzib made their first contribution in #58
• @lmmentel made their first contribution in #60
• @GuiArcencio made their first contribution in #66
• @scorcism made their first contribution in https://github.com/aeon-toolk...

sktime v0.16.0 (pre-fork)

Final pre-fork release. Subsequent releases will deviate from the original sktime repository, with the first release having breaking changes.

(see also https://github.com/sktime/sktime/releases/tag/v0.16.0)