spectral_connectivity.connectivity.MultivariateConnectivityResult#

class MultivariateConnectivityResult(method: str, scores: ndarray[tuple[int, ...], dtype[number]], connections: ndarray[tuple[int, ...], dtype[Any]], group_labels: ndarray[tuple[int, ...], dtype[Any]], group_membership: ndarray[tuple[int, ...], dtype[bool]], filters: ndarray[tuple[int, ...], dtype[floating]] | None = None, patterns: ndarray[tuple[int, ...], dtype[floating]] | None = None)[source]#

Bases: object

Component-resolved multivariate connectivity and spatial projections.

scores has shape (..., frequency, connection, component). Filters and patterns, when present, append (side, signal) where side 0 is the first group and side 1 is the second. Entries for signals outside a side’s group are NaN. connections contains the corresponding group-label pair for each connection and group_membership has shape (group, signal).

method#

Name of the measure that produced the result.

Type:: str

scores#

Per-component connectivity. Complex for canonical_coherency (magnitude times exp(-1j * phi)), real for MIC. A component a connection cannot supply (its smaller group has fewer channels than the requested n_components) is NaN.

Type:: NDArray[number], shape (…, frequency, connection, component)

connections#

The (first_group_label, second_group_label) pair for each connection.

Type:: NDArray, shape (connection, 2)

group_labels#

Sorted unique group labels.

Type:: NDArray, shape (group,)

group_membership#

True where a signal belongs to a group.

Type:: NDArray[bool], shape (group, signal)

filters#

Spatial filters mapping channel data to each component; NaN outside a side’s group.

Type:: NDArray[floating] or None, shape (…, frequency, connection, component, side, signal)

patterns#

Haufe-style patterns (within-group real CSD @ filter) mapping each component back to channel space.

Type:: NDArray[floating] or None, same shape as filters

Attributes:

filters
patterns

Methods

Attributes

`filters`
`patterns`
`method`
`scores`
`connections`
`group_labels`
`group_membership`