apca.metrics

Provided metrics for evaluating AugmentedPCA models.

Reconstruction Error

apca.metrics.reconstruct_error(a: numpy.ndarray, a_recon: numpy.ndarray, reduction: str = 'mean') → numpy.ndarray

Computes the reconstruction error between two matrices. This function can be used to determine the reconstruction error between original data matrices (primary or augmenting) and data matrices reconstructed via APCA.

Parameters

anumpy.ndarray

Original data matrix.

a_reconnumpy.ndarray

Reconstructed data matrix.

reductionstr; optional, default is ‘mean’

Specifies the reduction to apply to the output: ‘mean’ or ‘sum’. ‘mean’: mean of the squared error is taken, ‘sum’: sum of the squared error is taken.

Returns

errfloat

Reconstruction error between a and a_recon.