bins

often when analyzing data, I end up binning the data according to one variable and then, in each bin, computing some other quantity. this could be something as simple as a histogram or maybe more complicated. in order to bin your data, you need the bin edges; but to plot the data, you probably need the bin centers. rather than manually computing the midpoint of the edges (and remembering to adjust the calculation if your bins are logarithmically spaced), I use this snippet to make both the bin edges and centers in one go. also, I often find it useful to set the lowest bin edge to zero when the bins are logarithmically spaced, so that's included here too.

def make_bins(lo, hi, num, log=True, zero=True):
    space = np.geomspace if log else np.linspace
    all_bins = space(lo, hi, 2 * num + 1)
    edges = all_bins[0::2]
    centers = all_bins[1::2]
    if zero:
        edges[0] = 0
    return edges, centers

sample usage:

# data: x, y (ndarrays of same shape)

n_bins = 10
bin_e, bin_c = make_bins(1e-1, 1e2, n_bins, log=True, zero=True)

# use bin edges to compute per-bin quantities
y_means = np.zeros(n_bins)
for i in range(n_bins):
    mask = (x >= bin_e[i]) & (x < bin_e[i + 1])
    y_means[i] = np.average(y[mask])

# use bin centers to plot
plt.plot(bin_c, y_means)
plt.xscale('log')
plt.show()