Pyntbci

Added

- Added class `TRCA` to `transformers`
- Added class `eTRCA` to `classifiers`
- Added parameter `ensemble` to classes in `classifiers` to allow a separate spatial filter per class

Changed

- Changed package name from PyNT to PyntBCI to avoid clash with existing pynt library
- Changed filter order in `filterbank` of `utilities` to be optimized given input parameters

Fixed

- Fixed issue in `rCCA` of `classifiers` causing novel events in structure matrix when "cutting cycles"
- Fixed `correlation` to not contain mutable input variables

Added

- Added `tests`
- Added tutorials

Changed

- Changed `rCCA` to work with non-binary events instead of binary only

Fixed

Added

- Added dynamic stopping: classes `MarginStopping`, `BetaStopping`, and `BayesStopping` in module `stopping`
- Added value inner for variable `score_metric` in 'classifiers'

Changed

- Changed all data shapes from (channels, samples, trials) to (trials, channels, samples)
- Changed all codes shapes from (samples, classes) to (classes, samples)
- Changed all decision functions to similarity, not distance (e.g., Euclidean), to always maximize

Fixed

- Fixed zero-mean templates in `eCCA` and `rCCA` of `classifiers`

Added

- Added `Filterbank` to `classifiers`

Changed

- Changed classifiers all have `predict` and `decision_function` methods in `classifiers`

Fixed

Added

Changed

- Changed CCA method from sklearn to custom covariance method

Fixed

Added

- Added `eCCA` template metrics: average, median, OCSVM
- Added `eCCA` spatial filter options: all channels or subset

Changed

Fixed