v1.6.0

September 2026

28 people have contributed to this release through a collective 102 GitHub Pull Requests.

Highlights

• LearningShapeletClassifier has been removed, following its deprecation in v1.5.0.

• New estimators, including the ExtendedIsolationForest and MADRID anomaly detectors, the TimeSeriesAgglomerative clusterer and PrevalidatedRidgeClassifier.

• Significant speed-ups across many estimators and transforms, including HIVE-COTE, interval-based forests, rotation forest, BOSS and cBOSS, TDE, Arsenal and RandomShapeletTransform.

• Support for pandas 3, scikit-learn 1.9, and updates to many dependency bounds.

• Deprecated for removal in v1.7.0: the use_pycatch22 parameter, and the unused load_lynx, load_shampoo_sales and load_PBS_dataset dataset loaders.

• Lots of bug fixes!

Anomaly Detection

• [ENH] Changes left stampi so fit_is_empty (#3591) @TonyBagnall

• [BUG] Fix inverted MERLIN near-constant std check (#3763) @VedantMadane

• [BUG] reverse_windowing: make the psutil import optional (#3741) @mkzung

• [ENH] Add Extended Isolation Forest anomaly detector (#2113) (#3626) @CedricConday

• [ENH] Add MADRID multi-length discord anomaly detector (#3702) @JayeshSuryavanshi

Base

• [BUG] BaseCollectionEstimator: enforce the capability:univariate tag (#3742) @mkzung

• [MNT] Improve coverage for base package (#3516) @TonyBagnall

Classification

• [BUG] Relegate flat test to a warning (#3598) @TonyBagnall

• [BUG] Fix testing sklearn CV splitter to not have empty splits (#3608) @TonyBagnall

• [BUG] Fix MUSE with scikit-learn 1.8 and drop deprecated parameters (#3753) @MatthewMiddlehurst

• [BUG] Stabilize HIVE-COTE lifecycle test (#3762) @EltonChang1

• [BUG] Redcomet fix and testing (#3673) @TonyBagnall

• [MNT] Deprecate LearningShapeletClassifier (#3663) @TonyBagnall

• [MNT] dimensions -> channels (#3664) @TonyBagnall

• [DOC] finish remaining docstring clean-up in unequal_length (#3496) (#3686) @TonyBagnall

• [DOC] Fix class_weight description in numerous classifiers. (#3837) @johannfaouzi

• [ENH] Speed up interval-based forests (BaseIntervalForest / RandomIntervals) (#3620) @TonyBagnall

• [ENH] Speedup BOSS/cBOSS (up to 7x speedup, average 2x-3x) by vectorizing sparse matrix distance computations (#3640) @patrickzib

• [ENH] MiniRocketClassifier: change fitted estimator from protected to public (#3713) @montyy123

• [ENH] Rocket/MultiRocket/Hydra classifiers: change fitted estimators from protected to public (#3714) @montyy123

• [ENH] Speed up rotation forest (#3631) @TonyBagnall

• [ENH] Add PrevalidatedRidgeClassifier (#3670) @adam-goose

• [ENH] Replace the esig soft dependency with RoughPy (#3754) @MatthewMiddlehurst

• [ENH] Speed up TDE (#3625) @TonyBagnall

• [ENH] Speed up Arsenal classifier (#3655) @TonyBagnall

• [ENH] remove redundant Arsenal test for threaded output (#3829) @TonyBagnall

• [ENH] HC2 verbosity (#3644) @TonyBagnall

• [MNT] replace REDCOMETS imblearn oversampling with aeon SMOTE/ROS (#3669) @BB0813

• [MNT] Make estimator_ and transformers_ attributes public in BaseRIST (#3728) @aashoday

• [MNT] Make early_classification estimator_ and estimators_ attributes public (#3731) @aashoday

• [MNT] Make estimator and transformer attributes public in interval-based classifiers (#3733) @aashoday

• [MNT] Make classification/shapelet_based estimator_ and pipeline_ attributes public (#3730) @aashoday

• [MNT] Make estimators_ attribute public in HIVE-COTE classifiers (#3732) @aashoday

• [ENH] Speed up DrCIF part 1: Optimise Catch22 transformer (#3618) @TonyBagnall

• [BUG] return empty TIC outputs for all-stump forests (#3681) @AshSgDe29071999

Clustering

• [ENH] add time series agglomerative clusterer (#3553) @SotonSweetXss

• [ENH] Add capability:predict tag for Clustering (#3613) @TonyBagnall

• [BUG] KASBA fix and clustering tests (#3604) @TonyBagnall

• [MNT] Make TimeSeriesAgglomerative estimator_ attribute public (#3695) @aashoday

• [MNT] make clustering/feature_based estimator_ attributes public (#3703) @aashoday

• [MNT] make clustering/deep_learning estimator_ attributes public (#3718) @aashoday

Datasets

• [DEP] deprecate unused dataset loaders (#2835) (#3690) @TonyBagnall

• [DOC] Clear stale ImportError output from data_unequal notebook (#3748) @AnkitKumarIISERB

Dependencies

• [ENH] Replace the esig soft dependency with RoughPy (#3754) @MatthewMiddlehurst

• [MNT] Declare joblib and threadpoolctl as direct dependencies (#3785) @Obila34

• [MNT] Allow scikit 1.9 (#3787) @TonyBagnall

• [MNT] Update pandas upper bound to allow pandas 3 (#3834) @MatthewMiddlehurst

• [MNT] Update tslearn and sphinx upper bounds (#3835) @MatthewMiddlehurst

Distances

• [ENH] Remove dense bounding box from DTW (#3610) @TonyBagnall

Forecasting

• [BUG] Fix SCUM predict refitting (#3581) @TonyBagnall

• [BUG] NaiveForecaster: validate seasonal_period for strategy=’seasonal_last’ (#3615) @CedricConday

• [BUG] AutoARIMA: respect max_d in the differencing loop (#3614) @CedricConday

• [BUG] Fix iterative forecast when horizon >1 (#3667) @TonyBagnall

• [BUG] Fix numpy 2 incompatibilities in ClaSP, TRILITE and NBeats (#3795) @TonyBagnall

• [DOC] Rerun sast and forecasting notebooks to clear stale error outputs (#3813) @AnkitKumarIISERB

• [ENH] Better examples for CES and AutoCES (#3592) @GiGiKoneti

• [ENH] add drift strategy to NaiveForecaster (#3668) @TonyBagnall

• [MNT] Enhanced forecasting tests (#3580) @TonyBagnall

• [MNT] Forecasting test coverage (#3602) @TonyBagnall

• [MNT] Add CI test for correct API entries (#3750) @GiGiKoneti

Regression

• [BUG] Fix sporadic fail for Inception time file save (#3601) @TonyBagnall

• [ENH] Speed up interval-based forests (BaseIntervalForest / RandomIntervals) (#3620) @TonyBagnall

• [ENH] Speed up rotation forest (#3631) @TonyBagnall

• [ENH] Speed up interval-based forest regressors (#3628) @alexbanwell1

• [MNT] make regression estimator_ attributes public (#3712) @aashoday

• [MNT] Make estimator_ and transformers_ attributes public in BaseRIST (#3728) @aashoday

• [ENH] Speed up DrCIF part 1: Optimise Catch22 transformer (#3618) @TonyBagnall

Similarity Search

• [DOC] Use a shipped dataset in the SimHash index notebook (#3812) @L4XB

• [MNT] Similarity search cleanup (#3222) @baraline

• [MNT] Fix n_jobs > 3 in code speed notebook causing CI failures (#3639) @baraline

Transformations

• [BUG] Padder: validate fill_value and reject array-likes (#3495) (#3585) @CedricConday

• [BUG] Fix nightly tests related to SFA failing (#3612) @patrickzib

• [BUG] fix outlier_norm using wrong feature indices in pycatch22 path (#3638) @BB0813

• [BUG] fix uncommented njit decorator in sfa_fast (#3634) @patrickzib

• [BUG] fix ACF, STL and MSTL (#3651) @TonyBagnall

• [MNT] Improve transformations/series test coverage (#3533) @TonyBagnall

• [BUG] Fix PeriodogramTransformer crash on unequal-length input for non-constant pad_with (#3704) @Hrafz

• [BUG] DiscreteFourierApproximation: keep conjugate-symmetric terms (#3711) @Hrafz

• [BUG] ScaledLogitSeriesTransformer: apply the scaled logit when a bound is zero (#3740) @mkzung

• [BUG] CollectionToSeriesWrapper: return a series, not a collection of one (#3746) @mkzung

• [BUG] samplers: ask for the labels transform needs instead of failing on them (#3797) @mkzung

• [BUG] Fix numpy 2 incompatibilities in ClaSP, TRILITE and NBeats (#3795) @TonyBagnall

• [BUG] Redcomet fix and testing (#3673) @TonyBagnall

• [DOC] Fix BaseSeriesTransformer class docstring (#3801) @natsu0529

• [DOC] Rerun sast and forecasting notebooks to clear stale error outputs (#3813) @AnkitKumarIISERB

• [ENH] Improved coverage for channel selectors (#3522) @TonyBagnall

• [ENH] Speed up interval-based forests (BaseIntervalForest / RandomIntervals) (#3620) @TonyBagnall

• [ENH] Parallelize SAX and PAA transformers (#2980) @aadya940

• [ENH] speed up RandomShapeletTransform (#3574) @TonyBagnall

• [ENH] Enhancing the functionality of SAX (#3649) @hadifawaz1999

• [MNT] Deprecate pycatch22 (#3642) @TonyBagnall

• [ENH] Speed up interval-based forest regressors (#3628) @alexbanwell1

• [ENH] Replace the esig soft dependency with RoughPy (#3754) @MatthewMiddlehurst

• [ENH] Speed up Arsenal classifier (#3655) @TonyBagnall

• [ENH] Skip redundant per-slice input validation in interval feature extraction (#3665) @TonyBagnall

• [MNT] Deprecation of StatsModelsACF and StatsModelsPACF (#3653) @TonyBagnall

• [MNT] replace REDCOMETS imblearn oversampling with aeon SMOTE/ROS (#3669) @BB0813

• [MNT] Expose MSTLSeriesTransformer through public API (#3689) @Solaris-star

• [MNT] Expose YeoJohnsonTransformer through public API (#3739) @mkzung

• [MNT] Fix a typo in a warning message generated by RandomDilatedShapeletTransform (#3804) @johannfaouzi

• [DOC] Fix unequal length transformer docstrings (#3573) @kiwoongyoon

• [ENH] Speed up DrCIF part 1: Optimise Catch22 transformer (#3618) @TonyBagnall

Unit Testing

• [BUG] Use removesuffix(“.py”) when collecting changed files for numba cache invalidation (#3796) @Anai-Guo

• [ENH] Speed up interval-based forest regressors (#3628) @alexbanwell1

• [MNT] Enhanced forecasting tests (#3580) @TonyBagnall

• [MNT] Similarity search cleanup (#3222) @baraline

• [MNT] Fix check_non_state_changing_method (#3412) @MatthewMiddlehurst

• [MNT] Add CI test for correct API entries (#3750) @GiGiKoneti

Utilities

• [BUG] Remove unused broken compute_mean_stds_collection_parallel helper (#3734) @kiwoongyoon

• [ENH] Add parameters to method timer and stop fit timer from overwriting the attribute (#3656) @MatthewMiddlehurst

• [BUG] Znorm utils do not preserve float precision (#3707) @baraline

Visualisations

• [MNT] Use Agg backend in shapelet plotting tests (#3611) @TonyBagnall

• [BUG] Install graphviz for the notebook example runs (#3782) @TonyBagnall

Other

• [BUG] Uniformize kernelspace name and display_name across notebooks (#3636) @baraline

• [DOC] Update changelog.md for 1.5.0 and make requirement to edit this file more explicit (#3597) @TonyBagnall

• [DOC] Update funding information in about.md (#3660) @hadifawaz1999

• [DOC] Replace “indexes” with “indices” in docstrings for numpy consistency (#3736) @AnkitKumarIISERB

• [DOC] Add new papers to papers using aeon (#3786) @TonyBagnall

• [DOC] Fix wrong names in See Also sections (#3838) @johannfaouzi

• [MNT] Upgrade Python version to 3.13 in ReadTheDocs config (#3635) @MatthewMiddlehurst

Contributors

@aadya940, @aashoday, @adam-goose, @alexbanwell1, @Anai-Guo, @AnkitKumarIISERB, @AshSgDe29071999, @baraline, @BB0813, @CedricConday, @EltonChang1, @GiGiKoneti, @hadifawaz1999, @Hrafz, @JayeshSuryavanshi, @johannfaouzi, @kiwoongyoon, @L4XB, @MatthewMiddlehurst, @mkzung, @montyy123, @natsu0529, @Obila34, @patrickzib, @Solaris-star, @SotonSweetXss, @TonyBagnall, @VedantMadane