mvpa2.misc.neighborhood.CachedQueryEngine

class mvpa2.misc.neighborhood.CachedQueryEngine(queryengine)

Provides caching facility for query engines.

Notes

This QueryEngine simply remembers the results of the previous queries. Not much checking is done on either datasets it gets in train() is the same as the on in previous sweep of queries, i.e. either none of the relevant for underlying QueryEngine feature attributes was modified. So, CAUTION should be paid to avoid calling the same instance of CachedQueryEngine on different datasets (which might have different masking etc) .

query_byid() should be working reliably and without surprises.

query() relies on hashid of the queries, so there might be a collision! Thus consider it EXPERIMENTAL for now.

Attributes

queryengine

Methods

__call__(\*\*kwargs)
query(\*\*kwargs)	Return feature ids of neighbors given a specific query
query_byid(fid)	Return feature ids of neighbors for a given feature id
train(dataset)	‘Train’ CachedQueryEngine.
untrain()	Forgetting that CachedQueryEngine was already trained

Parameters:

queryengine : QueryEngine

Results of which engine to cache

Attributes

queryengine

Methods

__call__(\*\*kwargs)
query(\*\*kwargs)	Return feature ids of neighbors given a specific query
query_byid(fid)	Return feature ids of neighbors for a given feature id
train(dataset)	‘Train’ CachedQueryEngine.
untrain()	Forgetting that CachedQueryEngine was already trained

query(**kwargs)

Return feature ids of neighbors given a specific query

query_byid(fid)

Return feature ids of neighbors for a given feature id

queryengine

train(dataset)

‘Train’ CachedQueryEngine.

Raises:

ValueError

If dataset‘s .fa were changed – it would raise an exception telling to untrain explicitly, since the idea is to reuse CachedQueryEngine with the same engine and same dataset (up to variation of .sa, such as labels permutation)

untrain()

Forgetting that CachedQueryEngine was already trained