pyntbci.transformers.Vectorizer

class pyntbci.transformers.Vectorizer(channel_prime: bool = False)[source]

Vectorizer. Flattens a multi-dimensional data matrix per trial into a single feature vector, e.g. to use multi-channel time-series data with generic (non multi-dimensional) scikit-learn estimators.

Parameters:

channel_prime (bool (default: False)) – Whether the channels are the fastest-varying (i.e., contiguous) dimension in the flattened feature vector. If False, the samples are the fastest-varying dimension instead.

fit(X: NDArray, y: NDArray = None) → TransformerMixin[source]

Fit the vectorizer. Note, does not involve learning.

Parameters:

  • X (NDArray) – Data matrix of shape (n_trials, n_channels, n_samples).

  • y (NDArray (default: None)) – Not used.

Returns:

self – Returns the instance itself.

Return type:

TransformerMixin

transform(X: NDArray, y: NDArray = None) → NDArray[source]

Flatten the data matrix per trial into a single feature vector.

Parameters:

  • X (NDArray) – Data matrix of shape (n_trials, n_channels, n_samples).

  • y (NDArray (default: None)) – Not used.

Returns:

X – Flattened data matrix of shape (n_trials, n_channels * n_samples).

Return type:

NDArray