Ideally, you would also provide a full usage example in the docstring, including:
- An example call of the
COSgen CLI, with a short deliberation of the parameters you believe might most widely be varied by end users, and how they would choose how to set them (this sounds like much, but I guess it could be accomplished in less than one short paragraph per parameter). E.g.:
- does the HRF size count for the whole window? What if they were handed down a window with 50 TRs, where the last 20 TRs are at 0? is that a 30TR ot 50TR HRF?
- What can they expect to get - in simple terms from
autoregression=0.1 vs autoregression=0.9
- At what are more generations useless, and how can they diagnose this?
- Also, you could perhaps recommend that if the paradigm is new, they start with an estimation moel to actually determine their HRF (as we know optimal detection designs strongly depend on it).
- An example of how they can visualize the resulting sequence with the functionality COSgen provides.
Not least of all, I notice this CLI has no cutoff parameter. It would be very important to specify a lower bound (for a new estimation design, where the experimenter has good reason to fear that using single stimuli or even just too short blocks will render the measurement useless; or for detection designs, for reasons you already mentioned in the text).
Ideally, you would also provide a full usage example in the docstring, including:
COSgenCLI, with a short deliberation of the parameters you believe might most widely be varied by end users, and how they would choose how to set them (this sounds like much, but I guess it could be accomplished in less than one short paragraph per parameter). E.g.:autoregression=0.1vsautoregression=0.9Not least of all, I notice this CLI has no cutoff parameter. It would be very important to specify a lower bound (for a new estimation design, where the experimenter has good reason to fear that using single stimuli or even just too short blocks will render the measurement useless; or for detection designs, for reasons you already mentioned in the text).