[Numpy-discussion] an indexing question

Neal Becker ndbecker2 at gmail.com
Wed Jan 8 14:12:28 EST 2014

I have a 1d vector d.  I want compute the means of subsets of this vector.
The subsets are selected by looking at another vector s or same shape as d.

This can be done as:

[np.mean (d[s == i]) for i in range (size)]

But I think this could be done directly with numpy addressing, without resorting
to list comprehension?

More information about the NumPy-Discussion mailing list