Open thawn opened 1 year ago
Yeah, that's a good point. From my perspective, there are some functions where it might maybe make sense to implement SEM versions from. Also check documentation.
Tick all that apply for you:
Filters:
Graph based functions;
Projections:
Measurements:
We could then implement those similar to how variance and standard deviation are connected, e.g. here:; https://github.com/clEsperanto/pyclesperanto_prototype/blob/006379e7302cc0f76a571b5cdf3fb3962c9461c9/pyclesperanto_prototype/_tier2/_standard_deviation_box.py#L33-L35
I would keep it minimal if possible for now, and/or postpone implementing anything until we find a project where we need it. Also, e.g. standard-error-filter is not common afaik. But maybe it makes a lot of sense!
I cannot check the boxes, but I think SEM could be useful
I cannot check the boxes You should have received an invitation @thawn 👋
Some feature extraction functions return the standard deviation (STD). In some cases (i.e. when you want to know the accuracy of your measurement rather than the variance of your data) it may be useful to return the standard error of the mean (SEM) instead (or in addition)
SEM = STD/ sqrt(n) where n = number of measurements.