torch_ecg.utils.confusion_matrix

torch_ecg.utils.confusion_matrix(labels: ndarray | Tensor, outputs: ndarray | Tensor, num_classes: int | None = None) → ndarray

Compute a binary confusion matrix

The columns are ground truth labels and rows are predicted labels.

Parameters:

• labels (numpy.ndarray or torch.Tensor) – Binary labels, of shape (n_samples, n_classes), or indices of each label class, of shape (n_samples,).

• outputs (numpy.ndarray or torch.Tensor) – Binary outputs, of shape (n_samples, n_classes), or indices of each class predicted, of shape (n_samples,).

• num_classes (int, optional) – Number of classes. If labels and outputs are both of shape (n_samples,), then num_classes must be specified.

Returns:

cm – Confusion matrix, of shape (n_classes, n_classes).

Return type:

numpy.ndarray