"Observing histograms is annoying"

[axch]

Anthony said this when trying to reproduce the bird sexing by wingspan model that Chad Scherrer described in an introductory summer school lecture. I assume this means "From a histogram of results, produce an effect on the model equivalent to observing that a given nullary SP gave each result that many times."

Note: I view automatically dealing with binning a continuous output as out of scope -- assume the model SP is assessable all the way to the exact outputs specified in the histogram.

[lenaqr]

Yes, what I meant by "histogram" was just a vector of counts. (In this case I think it was actually a continuous variable that was binned, but that seemed to have been ignored in the treatment of the problem. I imagine the common use case of this for when the observed variable is discrete, such as a bag-of-words representation of a document.)