Marginalize entire column

[ibab]

It's useful to be able to integrate multi-dimensional distributions to marginalize out certain variables.
For example, when you want to compare a multi-dimensional distribution to data, you can histogram each column separately, and plot the distribution marginalized over all variables except the current one.
(When we leave out the other variables when plotting the data distribution, they essentially become "hidden" and need to be marginalized over when plotting the PDF).

Maybe something like

p = ... # multi-dimensional distribution
X = linspace(a, b, 200)
y = p.pdf(X, marginalize=(0, 2))
plt.plot(X, y)

could work.
Here, p would be three-dimensional and we would want to plot the second column.