example on numpy.histogram2d is broken (Trac #1830)

[numpy-gitbot]

Original ticket http://projects.scipy.org/numpy/ticket/1830 on 2011-05-14 by @samtygier, assigned to @pv.

the matplotlib example in http://docs.scipy.org/doc/numpy/reference/generated/numpy.histogram2d.html plots the data with the axis the wrong way. i think this is mostly the fault of imshow working in backwards coordinates.

the lines should be:

extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
plt.imshow(H.transpose(), extent=extent, interpolation='nearest', origin="lower")

test script (broken, as per numpy docs):

x = np.array([2,   2, -1, -1, -0.2])
y = np.array([1,-1.1,  1, -1,  0.5])
H, xedges, yedges = np.histogram2d(x,y, bins=(50, 50))
extent = [yedges[0], yedges[-1], xedges[-1], xedges[0]]
plt.imshow(H, extent=extent, interpolation='nearest')
plt.show()

test script (fixed):

x = np.array([2,   2, -1, -1, -0.2])
y = np.array([1,-1.1,  1, -1,  0.5])
H, xedges, yedges = np.histogram2d(x,y, bins=(50, 50))
extent = [xedges[0], xedges[-1], yedges[0], yedges[-1]]
plt.imshow(H.transpose(), extent=extent, interpolation='nearest', origin="lower")
plt.show()

[numpy-gitbot]

@rgommers wrote on 2011-07-06

There's a second documentation (or implementation) issue. From email by Alex Burton:

When normed=True the documentation says that each bin will be divided by its area,
whereas in implementation it is divided by area and the sum of all bins.

This should be checked for numpy 2.0; also that the implementation is consistent with the new kw for histogram. See #2224.

[charris]

The plot example was rewritten and looks fixed. The documentation issue originated in histogramdd, which also needs fixing.

[charris]

Note that the density keyword is missing from both histogram2d and histogramdd.