Hdlib

New features

- `examples/chopin2.py` now reports the Accuracy, F1, Precision, Recall, and the Matthews correlation coefficient for each of the folds in addition to the average of these scores as evaluation metrics of the hyperdimensional computing models;
- `model.Model` class functions now raise different exceptions based on multiple checks on the input parameters.

Fixes

- The `model.Model.stepwise_regression` function now report the importance corresponding to the best score;
- The `model.Model._init_fit_predict` function uses `average="weighted"` for computing a score different from the accuracy to account for label imbalance;
- `examples/chopin2.py` now computes different scores on the resulting predictions, prints the list of selected features based on the best score, and finally reports the confusion matrices.

New features

- Add `error_rate` as `model.Model` class method for computing the error rate of a classification model.

Fixes

- The `model.Model.predict` function computes the error rate before retraining the classification model.

Fixes

- Fix the retrining process in `model.Model.predict` to avoid overfitting.

:warning: Avoid using previous versions of `hdlib`.

Fixes

- Fix the initialization of Vector objects with a specific seed;
- `model.Model._init_fit_predict` and `model.Model._stepwise_regression_iter` are now private;
- Improving docstring adopting the [numpydoc](https://numpydoc.readthedocs.io/en/latest/) documentation format.

Fixes

- Fix the break condition in `model.Model.stepwise_regression` for both the `backward` and `forward` methods.