On Fri, Aug 27, 2010 at 14:05, Anne Archibald <aarchiba@physics.mcgill.ca> wrote:
On 27 August 2010 14:56, Robert Kern <robert.kern@gmail.com> wrote:
On Fri, Aug 27, 2010 at 13:38, <josef.pktd@gmail.com> wrote:
I don't think I have seen any higher dimensional kernel density estimation in python besides scipy.stats.kde. The Gaussian kde in scipy.stats is targeted to the underlying Fortran code for multivariate normal cdf.
Only for the "integrate over a box" functionality, which was what I needed at the time but is pretty rarely required otherwise. The rest is pure numpy.
I should say, integrating over a box is something I do all the time, though that is partly because it is cheap in my setting. For example, for plotting on a grid, what you really want to do is not sample on the grid but produce average values over the grid cells - this way you never miss or exaggerate a peak. So having efficient methods to integrate over one box or all grid cells can be really handy. Unfortunately I think it is often expensive even when approximations are made that allow discarding sufficiently distant points.
Well okay then. :-) -- Robert Kern "I have come to believe that the whole world is an enigma, a harmless enigma that is made terrible by our own mad attempt to interpret it as though it had an underlying truth." -- Umberto Eco