Sequentia

Major changes

- Switch from TravisCI to CircleCI. ([218](https://github.com/eonu/sequentia/issues/218))
- Add `datasets.load_random_sequences` for generating an arbitrarily sized dataset of sequences. ([216](https://github.com/eonu/sequentia/issues/216))
- Remove `DeepGRU` and `classifier.rnn` module. ([215](https://github.com/eonu/sequentia/issues/215))
- Add `sequentia.datasets` module. ([214](https://github.com/eonu/sequentia/issues/214))
- Added `return_scores` argument to `KNNClassifier.predict()` to return class scores. ([213](https://github.com/eonu/sequentia/issues/213))
- Return `self` in `fit()` functions. ([213](https://github.com/eonu/sequentia/issues/213))
- Update to `hmmlearn` v0.2.7. ([201](https://github.com/eonu/sequentia/issues/201))
- Update `HMMClassifier` structure to match `KNNClassifier`. ([200](https://github.com/eonu/sequentia/issues/200))
- Remove `'uniform'` `KNNClassifier` weighting option. ([192](https://github.com/eonu/sequentia/issues/192))
- Fix major `KNNClassifier` label scoring bug - thanks manisci. ([187](https://github.com/eonu/sequentia/issues/187))

Minor changes

- Update `CONTRIBUTING.md` CI instructions. ([219](https://github.com/eonu/sequentia/issues/219))
- Update HMM tests to use `datasets` module. ([217](https://github.com/eonu/sequentia/issues/217))
- Add `tslearn` as a core dependency. ([216](https://github.com/eonu/sequentia/issues/216))
- Remove `torchaudio`, `torchvision` and `torchfsdd` dependencies. ([214](https://github.com/eonu/sequentia/issues/214))
- Add playable audio to notebooks via `play_audio` helper. ([214](https://github.com/eonu/sequentia/issues/214))
- Update `README.md` and documentation. ([202](https://github.com/eonu/sequentia/issues/202))
- Add `Jinja2` dependency for RTD. ([188](https://github.com/eonu/sequentia/issues/188))

> - `KNNClassifier` has a major bug in all versions prior to and including v0.12.1 resulting in inaccurate predictions (see [186](https://github.com/eonu/sequentia/issues/186)).
> - `GMMHMM` and `HMMClassifier` have a major bug in all versions prior to and including v0.12.1 as a result of two bugs in the `GMMHMM` class in `hmmlearn` versions before v0.2.7 (see [193](https://github.com/eonu/sequentia/issues/193)).
>
> ⚠️ **Please use version v0.13.0 or later.**

Major changes

- Remove `requirements.py` due to import error. ([182](https://github.com/eonu/sequentia/pull/182))

Major changes

- Rework preprocessing module (see [177](https://github.com/eonu/sequentia/pull/177)). ([#179](https://github.com/eonu/sequentia/pull/179))
- Add `Custom` transformation.
- Rename `Preprocess` to `Compose`.
- Don't validate observation sequences after each transformation in `Compose`.
- Remove progress bars and `verbose` parameter.
- Stop unnecessarily copying each observation sequence before transformations.
- Change `transform()` function on `Transform` objects to accept a single observation sequence.
- Remove `_apply()` function on `Transform` objects.
- Make `_is_fitted()` public on `Transform` objects (change to `is_fitted()`).
- Use `__str__` instead of `_describe()` for transformation descriptions.
- Remove need to send `DeepGRU` to device explicitly, so we can now do `DeepGRU(..., device=device)` instead of `DeepGRU(..., device=device).to(device)`. ([178](https://github.com/eonu/sequentia/pull/178))
- Add `dev`, `test`, `docs` and `notebooks` extras. ([174](https://github.com/eonu/sequentia/pull/174))
- Remove `Equalize` transform as it goes against the point of variable-length sequence classification. ([172](https://github.com/eonu/sequentia/pull/172))
- Change `TrimZeros` transform to `TrimConstants`, allowing any constant-valued observation to be trimmed. ([172](https://github.com/eonu/sequentia/pull/172))
- Add DeepGRU classifier implementation. ([169](https://github.com/eonu/sequentia/pull/169))
- Add `sequentia[torch]` extra for optional `torch` CPU installation. ([169](https://github.com/eonu/sequentia/pull/169))

Minor changes

- Keep batch lengths on CPU ([pytorch/pytorch43227](https://github.com/pytorch/pytorch/issues/43227)). ([#178](https://github.com/eonu/sequentia/pull/178))
- Remove `docs/requirements.txt` and specify `docs` extra in `.readthedocs.yml`. ([176](https://github.com/eonu/sequentia/pull/176))
- Move Sphinx extensions from `docs/conf.py` to `requirements.py`. ([176](https://github.com/eonu/sequentia/pull/176))
- Bump development status classifier to beta. ([175](https://github.com/eonu/sequentia/pull/175))
- Move package dependency specifications to `requirements.py`. ([174](https://github.com/eonu/sequentia/pull/174))
- Add `docs/README.md`, `notebooks/README.md` and `lib/test/README.md`. ([174](https://github.com/eonu/sequentia/pull/174))
- Update HMM classifier diagram. ([173](https://github.com/eonu/sequentia/pull/173))
- Add build status to `README.md`. ([171](https://github.com/eonu/sequentia/pull/171))
- Fix patch description in `CONTRIBUTING.md`. ([170](https://github.com/eonu/sequentia/pull/170))
- Fix wording in `README.md`. ([167](https://github.com/eonu/sequentia/pull/167), [#168](https://github.com/eonu/sequentia/pull/168))

Major changes

- Fix validation for univariate sequences. ([164](https://github.com/eonu/sequentia/pull/164))

Minor changes

- Clean up `README.md` and add examples. ([165](https://github.com/eonu/sequentia/pull/165))
- Clean up validation logical expressions. ([164](https://github.com/eonu/sequentia/pull/164))

Major changes

- Add trailing underscore to variables containing trainable parameters (see 154). ([158](https://github.com/eonu/sequentia/pull/158))
- Add properties for GMM emission distribution parameters (see 153). ([156](https://github.com/eonu/sequentia/pull/156))
- Add selective `GMMHMM` parameter freezing/unfreezing (see 150). ([155](https://github.com/eonu/sequentia/pull/155))
- Fix random transition matrix initialization for `_LeftRightTopology` (see 149). ([151](https://github.com/eonu/sequentia/pull/151))

Minor changes

- Add access to Baum-Welch algorithm convergence monitor (see 139). ([162](https://github.com/eonu/sequentia/pull/162))
- Prefix `_Validator` functions with `is_` (see 159). ([161](https://github.com/eonu/sequentia/pull/161))
- Add validation for checking fitted parameters (see 157). ([160](https://github.com/eonu/sequentia/pull/160))
- Clean up `__repr__` for `GMMHMM`, `HMMClassifier` and `KNNClassifier`. ([160](https://github.com/eonu/sequentia/pull/160))
- Add classifier documentation links to `README.md`. ([152](https://github.com/eonu/sequentia/pull/152))
- Simplify random transition matrix initialization for `_LinearTopology` and `_LeftRightTopology`. ([151](https://github.com/eonu/sequentia/pull/151))

Major changes

- Fix `setup.py` encoding problem. ([145](https://github.com/eonu/sequentia/pull/145))
- Add `docs/robots.txt` and `sphinx-version-warning` package to prevent search engines from indexing old package versions (see 143). ([147](https://github.com/eonu/sequentia/pull/147))

Minor changes

- Add Prhmma as a contributor for 145. ([146](https://github.com/eonu/sequentia/pull/146))