Release 1.5.0 (#3571)

* Release 1.5.0

* ypdate changelog

* fix unusual test case

Author: TonyBagnall

aeon/__init__.py

@@ -1,3 +1,3 @@
 """aeon toolkit."""
 
-__version__ = "1.4.0"
+__version__ = "1.5.0"

aeon/clustering/deep_learning/tests/test_deep_clusterer_io.py

@@ -31,9 +31,13 @@
 @pytest.mark.parametrize("cls", ALL_DEEP_CLUSTERERS)
 def test_deep_clusterer_load_model(cls):
     """Test that all deep autoencoder clusterers load saved Keras models correctly."""
-    X = np.random.randn(4, 1, 10).astype(np.float32)
+    X = (
+        np.linspace(-1.0, 1.0, 10, dtype=np.float32)[None, None, :]
+        + np.arange(4, dtype=np.float32)[:, None, None]
+    )
     params = cls._get_test_params()[0]
     params["n_epochs"] = 1
+    params["random_state"] = 0
     params["save_best_model"] = True
 
     with tempfile.TemporaryDirectory() as tmp:

docs/changelogs/v1.5.md

@@ -0,0 +1,160 @@
+# v1.5.0
+
+June 2026
+
+31 people have contributed to this release through a collective 73 GitHub Pull
+Requests.
+
+## Highlights
+
+- New native implementations of estimators, including: forecasters NBeats, complex
+  exponentional smoothing (CES), dynamic optimised theta model (DOTM), a general
+  purpose forecasting ensemble and SCUM; TSelect channel selector;
+- New notebooks for the new forecasters
+- Documentation improvements and bug fixes
+
+## Anomaly Detection
+
+- [BUG] Rockad doctest fail fix ({pr}`3520`) {user}`Ton`yBagnall`
+
+## Classification
+
+- [ENH] Fix k-nn predict for k>1 ({pr}`3442`) {user}`TonyBagnall`
+- [BUG] ClassifierChannelEnsemble with remainder crashes ({pr}`3429`) {user}`SotonSweetXss`
+- [BUG] Fix predict_proba on unequal-length list input in Catch22 and TSFresh classifiers (#3427) ({pr}`3446`) {user}`gaoflow`
+- [BUG] Support callable weights in KNeighborsTimeSeriesClassifier ({pr}`3445`) {user}`gaoflow`
+- [DOC] Add Raises sections to BaseClassifier docstrings  ({pr}`3363`) {user} `Nandinisingh07`
+- [DOC] Fix classification docstring typos ({pr}`3443`) {user}`titanniya542-spec`
+- [ENH] Vectorize OOB divisor counting in RotationForest with np.bincount ({pr}`3457`) {user}`islem-esi`
+- [ENH] Deprecates LearningShapeletClassifier ({pr}`3485`) {user}`TonyBagnall`
+- [ENH] Refactor HIVE-COTE V1 and V2 to use a modular BaseHIVECOTE framework ({pr}`3386`) {user}`li-f8`
+- [ENH] remove prefer="threads" from Parallel in BaseIntervalForest ({pr}`3361`) {user}`TonyBagnall`
+- [BUG] use len(X) instead of X.shape[0] for list input in ShapeletTransformClassifier (fixes #3447) ({pr}`3465`) {user}`gaoflow`
+
+## Clustering
+
+- [BUG] Fix `TimeSeriesKernelKMeans` mutating `kernel` when `kernel="kdtw"` ({pr}`3424`) {user}`maxime2476`
+- [BUG] Validate uniqueness of custom medoid initialisation indices ({pr}`3430`) {user}`Booyaka101`
+- [BUG] Fix TimeSeriesCLARA sampling ({pr}`3456`) {user}`adam-goose`
+- [BUG] Reject array init for TimeSeriesCLARA ({pr}`3459`) {user}`ProfTrader`
+- [DOC] Fix clustering documentation typos ({pr}`3444`) {user}`SSDWGG`
+- [MNT] Deprecate TimeSeriesKShape wrapper from tslearn ({pr}`3381`) {user}`TonyBagnall`
+
+## Datasets
+
+- [BUG] Fix error in soft dependencies for HF hub in monster loader  ({pr}`3408`) {user}`Cyril-Meyer`
+- [BUG] Security issues for data loading ({pr}`3333`) {user}`MatthewMiddlehurst`
+- [BUG] Skip Monster dataset loader tests during PR CI (fixes #3466) ({pr}`3480`) {user}`gaoflow`
+- [BUG] Fix rehab loader  ({pr}`3521`) {user}`TonyBagnall`
+- [DEP] deprecation for 1.5 see #3440 ({pr}`3441`) {user}`TonyBagnall`
+
+## Distances
+
+- [DOC] Add Examples sections to ShapeDTW distance functions ({pr}`3486`) {user}
+  `laurapiro17`
+- [ENH] Add Dynamic Alphabet Sizes for SFA ({pr}`2844`) {user}`patrickzib`
+
+## Forecasting
+
+- [BUG] clone regressor in RegressionForecaster to prevent mutation (fixes #3448) ({pr}`3464`) {user}`gaoflow`
+- [BUG] Fix forecasting regression test: make _LastFeatureRegressor sklearn-clone-compatible ({pr}`3469`) {user}`TonyBagnall`
+- [BUG] Force ETS to fit seasonal when given period, ({pr}`3477`) {user}`TonyBagnall`
+- [BUG] Fixes AutoETS.iterative_forecast  ({pr}`3482`) {user}`TonyBagnall`
+- [ENH] NBeats forecaster added ({pr}`3357`) {user}`lucifer4073`
+- [ENH] Update iterative signature for theta ({pr}`3468`) {user}`TonyBagnall`
+- [ENH] Implement Complex Exponential Smoothing (CES) forecaster ({pr}`3463`) {user}`TonyBagnall`
+- [ENH] General purpose forecast ensemble ({pr}`3452`) {user}`TonyBagnall`
+- [ENH] Add noise to ets test ({pr}`3546`) {user}`TonyBagnall`
+- [ENH] Refactor iterative_forecast for fit/predict differentiation and exog support ({pr}`3454`) {user}`TonyBagnall`
+- [ENH] Fix ETS instability through bounds and limits to multiplicative ({pr}`3450`) {user}`TonyBagnall`
+- [ENH] Dynamic Optimised Theta Model (DOTM) forecaster ({pr}`3455`) {user}`TonyBagnall`
+- [ENH] Implement SCUM, the Simple Combination of Univariate Models forecaster ({pr}`3472`) {user}`TonyBagnall`
+- [DOC] Forecasting notebooks ({pr}`3550`) {user}`TonyBagnall`
+
+## Networks
+
+- [MNT] Apply precommit auto-fix from ruff ({pr}`3384`) {user}`SebastianSchmidl`
+- [MNT] Update `tensorflow` and refactor `networks` ({pr}`3308`) {user}`MatthewMiddlehurst`
+
+## Regression
+
+- [ENH] Vectorize OOB divisor counting in RotationForest with np.bincount ({pr}`3457`) {user}`islem-esi`
+
+## Similarity Search
+
+- [ENH] Improve coverage for similarity search ({pr}`3539`) {user}`TonyBagnall`
+
+## Transformations
+
+- [BUG] ShapletTransform contract bug and minor improvements ({pr}`3370`) {user}`TonyBagnall`
+- [BUG] Fix ESMOTE sample generation ({pr}`3504`) {user}`LinGinQiu`
+- [BUG] OHIT: seed sample generation and stop mutating k/kapa ({pr}`3505`) {user}`SAY-5`
+- [BUG] Stratify TSelect validation split ({pr}`3492`) {user}`nightcityblade`
+- [BUG] Respect requested distance in elbow class selector ({pr}`3493`) {user}`nightcityblade`
+- [BUG] Fix Padder add_noise for integer input ({pr}`3511`) {user}`gaoflow`
+- [BUG] Fix rocket random state ({pr}`3438`) {user}`TonyBagnall`
+- [BUG] Fix AutocorrelationFunctionTransformer crash with default n_lags=None ({pr}`3552`) {user}`glitch-ux`
+- [DOC] Fix imbalance transformer docstrings ({pr}`3509`) {user}`terminalchai`
+- [DOC] Fix docstring inaccuracies and typos in unequal_length package ({pr}`3528`) {user}`vir-cipher`
+- [DOC] Fix pad_length/truncated_length naming and test docstring in unequal_length ({pr}`3536`) {user}`gaurav-init`
+- [ENH] ShapeletTransform block transform for threads ({pr}`3372`) {user}`TonyBagnall`
+- [ENH]: add primitive type hints to LogTransformer ({pr}`3367`) {user}`Nandinisingh07`
+- [ENH]: add primitive type hints to BoxCoxTransformer - relates to #1454 ({pr}`3368`) {user}`Nandinisingh07`
+- [ENH] Enhanced verbose mode for RandomShapeletTransformer ({pr}`3383`) {user}`TonyBagnall`
+- [MNT] Improve test coverage for transformations/collection/dictionary_based (PAA, SAX, SFAWhole) ({pr}`3535`) {user}`TonyBagnall`
+- [ENH] Implement Tselect channel selection algorithm ({pr}`3402`) {user}`TonyBagnall`
+- [DOC] Fix docstring typos and inaccuracies in channel_selection (#3491) ({pr}`3530`){user}`MHJanny`
+- [BUG] fix(unequal_length): validate integer target lengths reject bool and non-positive values ({pr}`3507`) {user}`axelray-dev`
+- [BUG] fix: move RandomChannelSelector p validation from --init-- to _fit ({pr}`3541`) {user}`BB0813`
+
+## Other
+
+- [BUG] Restrict shapelet visualisation to 1D series ({pr}`3436`) {user}`TonyBagnall`
+- [BUG] Remove test exclusion for RDST #2290 ({pr}`3435`) {user}`TonyBagnall`
+- [MNT] Replace Flake8 with Ruff in pre-commit ({pr}`3161`) {user}`satwiksps`
+- [MNT] Add test to check that soft dependency skipping is working as intended ({pr}`3411`) {user}`MatthewMiddlehurst`
+- [BUG] fix soft deps ignoring PR_TESTING ({pr}`3404`) {user}`TonyBagnall`
+- [ENH] Add suggested fix field to bug template ({pr}`3556`) {user}`ProfTrader`
+- [DOC] Fix save_to_ts_file API reference (#3391) ({pr}`3393`) {user}`li-f8`
+- [DOC] Revised readme.md ({pr}`3388`) {user}`TonyBagnall`
+- [DOC] add HC2 to readme list ({pr}`3439`) {user}`TonyBagnall`
+- [DOC] Add automated mini-gallery examples for API pages ({pr}`3334`) {user}`Kaustbh`
+- [DOC] Update Code of Conduct reporting link to Monday form ({pr}`3478`) {user}`Copilot`
+- [MNT] change precommit yml ({pr}`3394`) {user}`TonyBagnall`
+- [MNT] Update `scikit-learn` upper bound ({pr}`3250`) {user}`MatthewMiddlehurst`
+
+
+## Contributors
+
+
+{user}`adam-goose`,
+{user}`axelray-dev`,
+{user}`BB0813`,
+{user}`Booyaka101`,
+{user}`Copilot`,
+{user}`Cyril-Meyer`,
+{user}`gaoflow`,
+{user}`gaurav-init`,
+{user}`glitch-ux`,
+{user}`islem-esi`,
+{user}`Kaustbh`,
+{user}`laurapiro17`,
+{user}`li-f8`,
+{user}`LinGinQiu`,
+{user}`lucifer4073`,
+{user}`MatthewMiddlehurst`,
+{user}`maxime2476`,
+{user}`MHJanny`,
+{user}`Nandinisingh07`,
+{user}`nightcityblade`,
+{user}`patrickzib`,
+{user}`ProfTrader`,
+{user}`satwiksps`,
+{user}`SAY-5`,
+{user}`SebastianSchmidl`,
+{user}`SotonSweetXss`,
+{user}`SSDWGG`,
+{user}`terminalchai`,
+{user}`titanniya542-spec`,
+{user}`TonyBagnall`,
+{user}`vir-cipher`

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "aeon"
-version = "1.4.0"
+version = "1.5.0"
 description = "A toolkit for time series machine learning"
 authors = [
     {name = "aeon developers", email = "contact@aeon-toolkit.org"},