Re: [Numpy-discussion] indexing with arrays of indexes and +=

[David H. <dav...@gm...>]

[snip]
Vincent,

I may be wrong, but it looks like you could use a weighted histogramdd (see
other thread where I sent a patch to do so).
You run it once to find the number of sample in each bin, then again with
weights=value, normed=False to find the total weight in each bin, then you
can divide each bin by the number of samples to find the average.

N, d = histogramdd(cells)
W, d = histogramdd(cells, weights = value)
average = W/N

David