Skip to content

Non default Sparsity kwargs when creating SortingAnalyzer #3320

@jonpedros

Description

@jonpedros

I'd like to create a SortingAnalyzer object with sparsity computed from channel groups rather than the default setting. However, it seems sparsity arguments passed to spikeinterface.core.create_sorting_analyzer() are passed to spikeinterface.core.estimate_sparsity(), which does not handle such method, instead of to spikeinterface.core.compute_sparsity().

At any rate, even with the relatively in-depth documentation specific to explaining sparsity, the many ways to generate it seem redundant/confusing. It seems compute_sparsity() is the most full-fledged function for generating it, but the fact it needs SortingAnalyzer to do so means one would have create SortingAnalyzer (possibly writing to disk etc) with sparse=False, generate sparsity and then create it (and write to disk) again.

Furthermore, it seems that default behavior is for the group property to be referred as 'group' for recordings, but when a grouped sorting is created from a grouped recording, this key changes automatically to 'channel_group', so compute_sparsity() will return an error when trying the routine I described above.

Metadata

Metadata

Assignees

No one assigned

    Labels

    questionGeneral question regarding SI

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions