classification

classification(
    U,
    Uhat,
    size=None,
    npc=None,
    metric='correlation',
    reduce_func=np.mean,
)

Movie time point/segment classification accuracy.

Parameters

Name	Type	Description	Default
U	(ndarray, shape(nt, nv))	The measured response patterns, often the columns are normalized PCs of the measured data rather than the original voxels/vertices.	required
Uhat	(ndarray, shape(nt, nv))	The predicted response patterns.	required
size	int or None	The number of time points in each segment. If None, each time point is treated as a segment.	`None`
npc	int or None	The number of PCs used for classification. If None, all PCs are used. If not None, assuming that `U` and `Uhat` are PCs, the first `npc` PCs are used for classification.	`None`
metric	str	The metric used for calculating the distance between two segments. See `scipy.spatial.distance.cdist` for the available metrics.	`'correlation'`
reduce_func	callable or None	The function used for reducing the classification accuracy across segments. If None, the raw classification accuracy is returned, where the length of the returned array is equal to the number of segments.	`np.mean`

Name	Type	Description
acc	float or ndarray	The classification accuracy. If `reduce_func` is None, the returned value is an ndarray of shape (nc,), where nc is the number of segments. Otherwise, the returned value is a float.