mvpa2.misc.surfing.queryengine.SurfaceQueryEngine

class mvpa2.misc.surfing.queryengine.SurfaceQueryEngine(surface, radius, distance_metric='dijkstra', fa_node_key='node_indices')

Query-engine that maps center nodes to indices of features (nodes) that are near each center node.

This class is for mappings from surface to surface features; for mappings from surface to voxel features, use SurfaceVerticesQueryEngine.

Methods

query(**kwargs)
query_byid(vertex_id)	Return feature ids of features near a vertex
train(ds)	Train the queryengine
untrain()

Make a new SurfaceQueryEngine

Parameters:

surface: surf.Surface or str :

surface object, or filename of a surface

radius: float :

size of neighborhood.

distance_metric: str :

‘euclidean’ or ‘dijkstra’ (default).

fa_node_key: str :

Key for feature attribute that contains node indices (default: ‘node_indices’).

Notes

After training this instance on a dataset and calling it with self.query_byid(vertex_id) as argument,

Methods

query(**kwargs)
query_byid(vertex_id)	Return feature ids of features near a vertex
train(ds)	Train the queryengine
untrain()

ids

query(**kwargs)

query_byid(vertex_id)

Return feature ids of features near a vertex

Parameters:

vertex_id: int :

Index of vertex (i.e. node) on the surface

Returns:

feature_ids: list of int :

Indices of features in the neighborhood of the vertex indexed by ‘vertex_id’

train(ds)

Train the queryengine

Parameters:

ds: Dataset :

dataset with surface data. It should have a field .fa.node_indices that indicates the node index of each feature.

untrain()