Sequentia

Major changes

- Clean up package imports. ([77](https://github.com/eonu/sequentia/pull/77))
- Rework `preprocessing` module. ([75](https://github.com/eonu/sequentia/pull/75))

Minor changes

- Fix typos and update preprocessing information in `README.md`. ([76](https://github.com/eonu/sequentia/pull/76))

Major changes

- Remove strict requirement of Numpy arrays being two-dimensional by using `numpy.atleast_2d` to convert one-dimensional arrays into 2D. ([70](https://github.com/eonu/sequentia/pull/70))

Minor changes

- As the HMM classifier is not a true ensemble of HMMs (since each HMM doesn't really contribute to the classification), it is no longer referred to as an ensemble. ([69](https://github.com/eonu/sequentia/pull/69))

Major changes

- Add package tests and Travis CI support. ([56](https://github.com/eonu/sequentia/pull/56))
- Remove Python v3.8+ support. ([56](https://github.com/eonu/sequentia/pull/56))
- Rename `normalize` preprocessing method to `center`, since it just centers an observation sequence. ([62](https://github.com/eonu/sequentia/pull/62))
- Add `standardize` preprocessing method for standardizing (standard scaling) an observation sequence. ([63](https://github.com/eonu/sequentia/pull/63))
- Add `trim_zeros` preprocessing method for removing zero-observations from an observation sequence. ([67](https://github.com/eonu/sequentia/pull/67))

Minor changes

- Add `Validator.random_state` for validating random state objects and seeds. ([56](https://github.com/eonu/sequentia/pull/56))
- Internalize `Validator` and topology (`Topology`, `ErgodicTopology`, `LeftRightTopology`) classes. ([57](https://github.com/eonu/sequentia/pull/57))
- Use proper documentation format for topology classes. ([58](https://github.com/eonu/sequentia/pull/58))

Major changes

- Add `Preprocess.summary()` to display an ordered summary of preprocessing transformations. ([54](https://github.com/eonu/sequentia/pull/54))
- Add mean and median filtering preprocessing methods. ([48](https://github.com/eonu/sequentia/pull/48))
- Use median filtering and decimation downsampling by default. ([52](https://github.com/eonu/sequentia/pull/52))
- Modify preprocessing boundary conditions ([51](https://github.com/eonu/sequentia/pull/51)):
- Use a bi-directional window for filtering to resolve boundary problems.
- Modify downsampling method to downsample residual observations.

Minor changes

- Add supported topologies (left-right and ergodic) to feature list. ([53](https://github.com/eonu/sequentia/pull/53))
- Add restrictions on preprocessing parameters: downsample factor and window size. ([50](https://github.com/eonu/sequentia/pull/50))
- Allow `Preprocess` class to be used to apply preprocessing transformations to a single observation sequence. ([49](https://github.com/eonu/sequentia/pull/49))

Major changes

- Re-add `euclidean` metric as `DTWKNN` default. ([43](https://github.com/eonu/sequentia/pull/43))

Minor changes

- Add explicit labels to `evaluate()` in `HMMClassifier` example. ([44](https://github.com/eonu/sequentia/pull/44))

Major changes

- Add proper documentation, hosted on [Read The Docs](https://sequentia.readthedocs.io/en/latest). ([#40](https://github.com/eonu/sequentia/pull/40), [#41](https://github.com/eonu/sequentia/pull/41))