Releases: awslabs/gluonts
Releases · awslabs/gluonts
0.15.1
0.15.0
New features
- add iTransformer multivariate forecaster (#3017) by @kashif
- Add TiDE model (#3096) by @maxc01
- Add ercot dataset (#3148) by @lostella
- Add ETT datasets (#3149) by @lostella
- Add item_id to ERCOT and ETT datasets (#3150) by @shchur
- Add Seasonal Aggregate Predictor (#3162) by @rshyamsundar
Model updates
- Add option to provide function for season_length in SeasonalNaivePredictor (#3033) by @lostella
- R: add ARIwith exogenous regressors (#3086) by @leica2023
Breaking changes
- ev: Batch data for faster evaluation. (#3051) by @lostella
- Unify QuantileOutput and DistributionOutput (#3093) by @shchur
- Rotbaum: turn to json-based serialization (#3176) by @lostella
Fixes
- Fix loader for the M5 dataset (#3151) by @shchur
- Fix loaders for M5 & ETT datasets (#3155) by @shchur
- Fix item_id for M5 dataset (#3156) by @shchur
- Serde: limit decode code execution (#3175) by @lostella
Dependencies
- Nursery: bump up pyarrow requirement in gluonts.nursery.tsbench (#3056) by @lostella
- Bump scikit-learn from 0.23.2 to 1.0.1 in /src/gluonts/nursery/daf (#3118) by @dependabot
- Bump sklearn version in nursery subpackages (#3120) by @lostella
- Bump dependencies versions to address dependabot alerts (#3173) by @lostella
- Bump dependencies versions to address dependabot alerts, again (#3174) by @lostella
Other changes
- update dockerfile python version to 3.8 (#3072) by @melopeo
- Update REFERENCES.md (#3078) by @abdulfatir
- CI: run R tests on push (#3089) by @lostella
- CI: chain requirements to ensure compatibility (#3112) by @Borda
- test: unify using @pytest.mark.flaky(...) & bump pytest plugins (#3132) by @Borda
- Add Chronos Breaking News (#3154) by @abdulfatir
- lint: freeze & run Black version 24.02 (#3131) by @Borda
- ci: update lints Ruff & docformatter (#3130) by @Borda
- Extend test coverage for pandas frequencies (#3179) by @shchur
- Tests: xfail test for fourier.arima.xreg (#3182) by @lostella
0.14.4
0.14.3
Backporting:
- Fix Rotbaum serialization and deserialization #3068 by @pantanurag555
- Fix Rotbaum to handle short series #3073 by @leica2023
0.13.9
Backporting:
- Fix Rotbaum serialization and deserialization #3068 by @pantanurag555
- Fix Rotbaum to handle short series #3073 by @leica2023
0.14.2
0.14.1
0.13.8
0.14.0
See full diff here: v0.13.0...v0.14.0
Breaking changes
- Simplify plotting of forecasts. #2864
- Breaking: change num_output with quantiles in TFT (mxnet) #2879
- DeepVARHierarchicalEstimator: Remove target_dim parameter. #2889
- Rework maybe.Maybe to use a boxed container. #2900
- Rework ev. #2914
- gluonts.ev: Rework exports. #2928
- Remove freq argument in SeasonalNaivePredictor #2932
- Remove freq argument from naive2 #2935
- Simplify PyTorchPredictor serde #2965
- Move QuantizeScaled from Wavenet to gluonts.transform, rename to QuantizeMeanScaled #2975
- Change torch state_dict file name to prediction-net-state.pt #2985
- Remove freq attribute from NPTS #3009
0.14.0 rc2
Backports for v0.14.0rc2 (#3032) * Fix WaveNet inputs (#3022) * Fix version range for lightning (#3023) * Allow VCS install using GLUONTS_FALLBACK_VERSION (#3028) * Make `mypy` checks opt-out instead of opt-in, fix several type issues (#3027) Co-authored-by: Oleksandr Shchur <shchuro@amazon.com> * No API docs for nursery. (#3030) * Fix: #3030. (#3031) * Add support for Pydantic v1 and v2. (#3026) --------- Co-authored-by: Oleksandr Shchur <shchuro@amazon.com> Co-authored-by: ddelange <14880945+ddelange@users.noreply.github.com> Co-authored-by: Jasper <schjaspe@amazon.de>