v0.3.0#

June 2023

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

Benchmarking#

Enhancements#

[ENH] Implement simple(r) benchmarking (#379) @DBCerigo
[ENH] remove unsupported legacy benchmarking code (#439) @TonyBagnall
[ENH] Rework datasets package notebooks (#468) @TonyBagnall

Classification#

Bug Fixes#

[BUG] refactored params usage resnet (#433) @hadifawaz1999
[BUG] Ordinal TDE _n_jobs (#471) @MatthewMiddlehurst

Enhancements#

[ENH] Parametrize model saving for deep classifiers and regressors (#430) @hadifawaz1999
[ENH] Ordinal TDE algorithm and extension of SFA transformer (#335) @RafaAyGar
[ENH] Facilitate multivariate and unequal length with ElasticEnsemble classifier (#415) @TonyBagnall
[ENH] refactor _threads_to_use to _n_jobs (#377) @TonyBagnall
[ENH] MrSQM classification wrapper (#476) @MatthewMiddlehurst

Refactored#

[ENH] refactor _threads_to_use to _n_jobs (#377) @TonyBagnall

Other#

[ENH] Allow KNN to take unequal length series. (#412) @TonyBagnall

Clustering#

Bug Fixes#

[BUG] Tslearn clusterers update (#457) @chrisholder

Distances#

Bug Fixes#

[BUG] KNN distance params (#447) @chrisholder

Enhancements#

[ENH] updated dba to support custom parameters (#454) @chrisholder

Other#

[ENH] Allow KNN to take unequal length series. (#412) @TonyBagnall

Regression#

Enhancements#

[ENH] Parametrize model saving for deep classifiers and regressors (#430) @hadifawaz1999
[ENH] Enable knn regressor to take unequal length (#442) @TonyBagnall
[MNT] Add deep learning regression/classification test (#443) @hadifawaz1999

Transformations#

Bug Fixes#

[ENH] Convert IntervalSegmenter and RandomIntervalSegmenter to use numpy3D internally (#391) @TonyBagnall
[BUG] Fixes RDST bug and small docstring changes (#475) @MatthewMiddlehurst

Enhancements#

[ENH] Ordinal TDE algorithm and extension of SFA transformer (#335) @RafaAyGar
[ENH] Collection transformer base class (#263) @MatthewMiddlehurst
[ENH] Refactor sax (#417) @hadifawaz1999

Maintenance#

[MNT] Rename panel transformers folder to collection (#466) @MatthewMiddlehurst

Refactored#

[ENH] Refactor sax (#417) @hadifawaz1999
[MNT] Rename panel transformers folder to collection (#466) @MatthewMiddlehurst

Other#

Bug Fixes#

[BUG] plot_series bugfix and tests (#318) @MatthewMiddlehurst
[BUG] Fixes stratified_resample so that it works with 3D numpy (#460) @GuiArcencio

Documentation#

[DOC] Add to getting_started page (#393) @TonyBagnall
[DOC] README deprecation policy disclaimer (#453) @MatthewMiddlehurst
[DOC] README getting started section (#462) @MatthewMiddlehurst
[DOC] Getting started page pipeline examples (#461) @MatthewMiddlehurst

Enhancements#

[ENH] numba utility functions (#465) @MatthewMiddlehurst
[ENH] data loaders and writers (#463) @TonyBagnall

Maintenance#

[MNT] Set upper bound on holidays to fix CI (#436) @MatthewMiddlehurst
[MNT] Recursively find notebooks for testing (#470) @MatthewMiddlehurst

Other#

[EN] Hard code deep learning exclusions from tests (#419) @TonyBagnall
[ENH] Remove the uses of the nested univariate data generator (#373) @TonyBagnall

Contributors#

The following have contributed to this release through a collective 33 GitHub Pull Requests:

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