BEP041: Statistical Model Derivatives

[tsalo]

Some folks (myself included) are interested in adding common imaging model derivatives (especially GLM outputs) to the specification. fitlins currently writes out files with a number of entities and suffixes that are not yet part of the specification. As such, I have started drafting a BEP that would explicitly add these entities and suffixes to the specification at https://docs.google.com/document/d/1KHzp-yk8KXvkUIhtN71WU0m4P4kKT9C1yvI-i9_kNeY/edit?usp=sharing.

This issue stems from poldracklab/fitlins#299.

[yarikoptic]

FWIW, ran into this issue while looking for "model" entity issue, model is being added within bep039: Dimensionality reduction-based networks . So here you mean "stats-model" @tsalo ? also attn @PeerHerholz

[PeerHerholz]

Ahoi hoi @yarikoptic (& @tsalo),

the "model" entity issues is/was being discussed here.
Also, model- is potentially being added via BEP16 before.

Cheers, Peer

[tsalo]

@yarikoptic right, I'm referring to stats models (primarily GLMs). There might be some overlap with BEP16, but it doesn't seem substantial.

[yarikoptic]

My comment was primarily about the fact that the entity model is being introduced within a somewhat niche for that semantic thus possibly later causing "conflicts" or confusions with other BEPs etc.

[effigies]

@tsalo We should give this BEP a number and add it to the website, so it's findable.

[tsalo]

Great idea. I've opened bids-standard/bids-website#343.