|
2 | 2 | # Changelog |
3 | 3 |
|
4 | 4 | We do our best to avoid the introduction of breaking changes, |
5 | | -but cannot always guarantee backwards compatibility. Changes that may **break code which uses a previous release of Darts** are marked with a "🔴". |
| 5 | +but cannot always guarantee backwards compatibility. Changes that may **break code which uses a previous release of Darts** are marked with a "🔴". |
6 | 6 |
|
7 | 7 | ## [Unreleased](https://github.com/unit8co/darts/tree/master) |
8 | | -- Created `ShapExplainabilityResult` by extending `ExplainabilityResult`. This subclass carries additional information |
9 | | - specific to Shap Explainers (i.e., the corresponding feature values and the underlying `shap.Explanation` object). |
10 | | - [#1545](https://github.com/unit8co/darts/pull/1545) by [Rijk van der Meulen](https://github.com/rijkvandermeulen). |
11 | | - |
12 | | -[Full Changelog](https://github.com/unit8co/darts/compare/0.23.1...master) |
13 | | -- `LightGBM` model now supports native categorical feature handling as described |
14 | | - [here](https://lightgbm.readthedocs.io/en/latest/Features.html#optimal-split-for-categorical-features). |
15 | | - [#1585](https://github.com/unit8co/darts/pull/1585) by [Rijk van der Meulen](https://github.com/rijkvandermeulen) |
| 8 | +[Full Changelog](https://github.com/unit8co/darts/compare/0.24.0...master) |
| 9 | + |
| 10 | +## [0.24.0](https://github.com/unit8co/darts/tree/0.24.0) (2023-04-12) |
| 11 | +### For users of the library: |
| 12 | + |
| 13 | +**Improved** |
| 14 | +- General model improvements: |
| 15 | + - New baseline forecasting model `NaiveMovingAverage`. [#1557](https://github.com/unit8co/darts/pull/1557) by [Janek Fidor](https://github.com/JanFidor). |
| 16 | + - New models `StatsForecastAutoCES`, and `StatsForecastAutoTheta` from Nixtla's statsforecasts library as local forecasting models without covariates support. AutoTheta supports probabilistic forecasts. [#1476](https://github.com/unit8co/darts/pull/1476) by [Boyd Biersteker](https://github.com/Beerstabr). |
| 17 | + - Added support for future covariates, and probabilistic forecasts to `StatsForecastAutoETS`. [#1476](https://github.com/unit8co/darts/pull/1476) by [Boyd Biersteker](https://github.com/Beerstabr). |
| 18 | + - Added support for logistic growth to `Prophet` with parameters `growth`, `cap`, `floor`. [#1419](https://github.com/unit8co/darts/pull/1419) by [David Kleindienst](https://github.com/DavidKleindienst). |
| 19 | + - Improved the model string / object representation style similar to scikit-learn models. [#1590](https://github.com/unit8co/darts/pull/1590) by [Janek Fidor](https://github.com/JanFidor). |
| 20 | + - 🔴 Renamed `MovingAverage` to `MovingAverageFilter` to avoid confusion with new `NaiveMovingAverage` model. [#1557](https://github.com/unit8co/darts/pull/1557) by [Janek Fidor](https://github.com/JanFidor). |
| 21 | +- Improvements to `RegressionModel`: |
| 22 | + - Optimized lagged data creation for fit/predict sets achieving a drastic speed-up. [#1399](https://github.com/unit8co/darts/pull/1399) by [Matt Bilton](https://github.com/mabilton). |
| 23 | + - Added support for categorical past/future/static covariates to `LightGBMModel` with model creation parameters `categorical_*_covariates`. [#1585](https://github.com/unit8co/darts/pull/1585) by [Rijk van der Meulen](https://github.com/rijkvandermeulen). |
| 24 | + - Added lagged feature names for better interpretability; accessible with model property `lagged_feature_names`. [#1679](https://github.com/unit8co/darts/pull/1679) by [Antoine Madrona](https://github.com/madtoinou). |
| 25 | + - 🔴 New `use_static_covariates` option for all models: When True (default), models use static covariates if available at fitting time and enforce identical static covariate shapes across all target `series` used for training or prediction; when False, models ignore static covariates. [#1700](https://github.com/unit8co/darts/pull/1700) by [Dennis Bader](https://github.com/dennisbader). |
| 26 | +- Improvements to `TorchForecastingModel`: |
| 27 | + - New methods `load_weights()` and `load_weights_from_checkpoint()` for loading only the weights from a manually saved model or checkpoint. This allows to fine-tune the pre-trained models with different optimizers or learning rate schedulers. [#1501](https://github.com/unit8co/darts/pull/1501) by [Antoine Madrona](https://github.com/madtoinou). |
| 28 | + - New method `lr_find()` that helps to find a good initial learning rate for your forecasting problem. [#1609](https://github.com/unit8co/darts/pull/1609) by [Levente Szabados](https://github.com/solalatus) and [Dennis Bader](https://github.com/dennisbader). |
| 29 | + - Improved the [user guide](https://unit8co.github.io/darts/userguide/torch_forecasting_models.html) and added new sections about saving/loading (checkpoints, manual save/load, loading weights only), and callbacks. [#1661](https://github.com/unit8co/darts/pull/1661) by [Antoine Madrona](https://github.com/madtoinou). |
| 30 | + - 🔴 Replaced `":"` in save file names with `"_"` to avoid issues on some operating systems. For loading models saved on earlier Darts versions, try to rename the file names by replacing `":"` with `"_"`. [#1501](https://github.com/unit8co/darts/pull/1501) by [Antoine Madrona](https://github.com/madtoinou). |
| 31 | + - 🔴 New `use_static_covariates` option for `TFTModel`, `DLinearModel` and `NLinearModel`: When True (default), models use static covariates if available at fitting time and enforce identical static covariate shapes across all target `series` used for training or prediction; when False, models ignore static covariates. [#1700](https://github.com/unit8co/darts/pull/1700) by [Dennis Bader](https://github.com/dennisbader). |
| 32 | +- Improvements to `TimeSeries`: |
| 33 | + - Added support for integer indexed input to `from_*` factory methods, if index can be converted to a pandas.RangeIndex. [#1527](https://github.com/unit8co/darts/pull/1527) by [Dennis Bader](https://github.com/dennisbader). |
| 34 | + - Added support for integer indexed input with step sizes (freq) other than 1. [#1527](https://github.com/unit8co/darts/pull/1527) by [Dennis Bader](https://github.com/dennisbader). |
| 35 | + - Optimized time series creation with `fill_missing_dates=True` achieving a drastic speed-up . [#1527](https://github.com/unit8co/darts/pull/1527) by [Dennis Bader](https://github.com/dennisbader). |
| 36 | + - `from_group_dataframe()` now warns the user if there is suspicion of a "bad" time index (monotonically increasing). [#1628](https://github.com/unit8co/darts/pull/1628) by [Dennis Bader](https://github.com/dennisbader). |
| 37 | +- Added a parameter to give a custom function name to the transformed output of `WindowTransformer`; improved the explanation of the `window` parameter. [#1676](https://github.com/unit8co/darts/pull/1676) and [#1666](https://github.com/unit8co/darts/pull/1666) by [Jing Qiang Goh](https://github.com/JQGoh). |
| 38 | +- Added `historical_forecasts` parameter to `backtest()` that allows to use precomputed historical forecasts from `historical_forecasts()`. [#1597](https://github.com/unit8co/darts/pull/1597) by [Janek Fidor](https://github.com/JanFidor). |
| 39 | +- Added feature values and SHAP object to `ShapExplainabilityResult`, giving easy user access to all SHAP-specific explainability results. [#1545](https://github.com/unit8co/darts/pull/1545) by [Rijk van der Meulen](https://github.com/rijkvandermeulen). |
| 40 | +- New `quantile_loss()` (pinball loss) metric for probabilistic forecasts. [#1559](https://github.com/unit8co/darts/pull/1559) by [Janek Fidor](https://github.com/JanFidor). |
| 41 | + |
| 42 | +**Fixed** |
| 43 | +- Fixed an issue in `BottomUp/TopDownReconciliator` where the order of the series components was not taken into account. [#1592](https://github.com/unit8co/darts/pull/1592) by [David Kleindienst](https://github.com/DavidKleindienst). |
| 44 | +- Fixed an issue with `DLinearModel` not supporting even numbered `kernel_size`. [#1695](https://github.com/unit8co/darts/pull/1695) by [Antoine Madrona](https://github.com/madtoinou). |
| 45 | +- Fixed an issue with `RegressionEnsembleModel` not using future covariates during training. [#1660](https://github.com/unit8co/darts/pull/1660) by [Rajesh Balakrishnan](https://github.com/Rajesh4AI). |
| 46 | +- Fixed an issue where `NaiveEnsembleModel` prediction did not transfer the series' component name. [#1602](https://github.com/unit8co/darts/pull/1602) by [David Kleindienst](https://github.com/DavidKleindienst). |
| 47 | +- Fixed an issue in `TorchForecastingModel` that prevented from using multi GPU training. [#1509](https://github.com/unit8co/darts/pull/1509) by [Levente Szabados](https://github.com/solalatus). |
| 48 | +- Fixed a bug when saving a `FFT` model with `trend=None`. [#1594](https://github.com/unit8co/darts/pull/1594) by [Antoine Madrona](https://github.com/madtoinou). |
| 49 | +- Fixed some issues with PyTorch-Lightning version 2.0.0. [#1651](https://github.com/unit8co/darts/pull/1651) by [Dennis Bader](https://github.com/dennisbader). |
| 50 | +- Fixed a bug in `QuantileDetector` which raised an error when low and high quantiles had identical values. [#1553](https://github.com/unit8co/darts/pull/1553) by [Julien Adda](https://github.com/julien12234). |
| 51 | +- Fixed an issue preventing `TimeSeries` from being empty. [#1359](https://github.com/unit8co/darts/pull/1359) by [Antoine Madrona](https://github.com/madtoinou). |
| 52 | +- Fixed an issue when using `backtest()` on multiple series. [#1517](https://github.com/unit8co/darts/pull/1517) by [Julien Herzen](https://github.com/hrzn). |
| 53 | +- General fixes to `historical_forecasts()` |
| 54 | + - Fixed issue where `retrain` functions were not handled properly; Improved handling of `start`, and `train_length` parameters; better interpretability with warnings and improved error messages (warnings can be turned of with `show_warnings=False`). By [#1675](https://github.com/unit8co/darts/pull/1675) by [Antoine Madrona](https://github.com/madtoinou) and [Dennis Bader](https://github.com/dennisbader). |
| 55 | + - Fixed an issue for several models (mainly ensemble and local models) where automatic `start` did not respect the minimum required training lengths. [#1616](https://github.com/unit8co/darts/pull/1616) by [Janek Fidor](https://github.com/JanFidor) and [Dennis Bader](https://github.com/dennisbader). |
| 56 | + - Fixed an issue when using a `RegressionModel` with future covariates lags only. [#1685](https://github.com/unit8co/darts/pull/1685) by [Maxime Dumonal](https://github.com/dumjax). |
| 57 | + |
| 58 | +### For developers of the library: |
| 59 | + |
| 60 | +**Improvements** |
| 61 | +- Option to skip slow tests locally with `pytest . --no-cov -m "not slow"`. [#1625](https://github.com/unit8co/darts/pull/1625) by [Blazej Nowicki](https://github.com/BlazejNowicki). |
| 62 | +- Major refactor of data transformers which simplifies implementation of new transformers. [#1409](https://github.com/unit8co/darts/pull/1409) by [Matt Bilton](https://github.com/mabilton). |
| 63 | + |
16 | 64 |
|
17 | 65 | ## [0.23.1](https://github.com/unit8co/darts/tree/0.23.1) (2023-01-12) |
18 | 66 | Patch release |
|
0 commit comments