matplotlib: dual chart (plot and events)

I like @jcoppens idea with axhspan. Here is an example using bar on top and step for the main plot. The top bar chart corresponds to the orange step plot on the bottom, and similarly for the green bar chart.

The orange or green color map in the bar charts corresponds to the number of people coming in (determined by a positive value in the associated step plot - by associated I mean green with green, orange with orange). The grayscale corresponds to people going out (determined by negative values in the associated step plot).

Code to generate the plot:

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.cm as cm
import matplotlib.colors as col

hr = ['{num:02d}'.format(num=i) for i in range(24)]
times = ['{0}:00'.format(h) for h in hr ]
tick_location = [t*4 for t in range(24)]

x1 = np.random.randint(-10, 20, 4*24)
x1[0:5] = 0
x1[20:30] = 0
x1[42:57] = 0
x1[69:74] = 0
x2 = np.random.randint(-3, 5, 4*24)
x2[0:25] = 0
x2[48:57] = 0
x2[89:] = 0

What you actually care about:

cmap_in_x1 = cm.ScalarMappable(col.Normalize(min(x1), max(x1)), 'Greens')
cmap_in_x2 = cm.ScalarMappable(col.Normalize(min(x2), max(x2)), 'Oranges')
cmap_out = cm.ScalarMappable(col.Normalize(min(x1), max(x1)), 'bone')

fig = plt.figure(figsize=(12,8))

big_ax = fig.add_axes([0.1, 0.1, 0.8, 0.6])
plt.grid()

small_ax_1 = fig.add_axes([0.1, 0.7, 0.8, 0.05], sharex=big_ax)
small_ax_2 = fig.add_axes([0.1, 0.75, 0.8, 0.05], sharex=big_ax)

big_ax.step(np.arange(len(x1)), x1, color='lightgreen', alpha=0.8, linewidth=2)
big_ax.step(np.arange(len(x2)), x2, color='orange', alpha=0.5, linewidth=2)

small_ax_1.bar(np.arange(len(x1>0)), x1>0, align='center', alpha=0.8, color=cmap_in_x1.to_rgba(x1))
small_ax_1.bar(np.arange(len(x1<0)), x1<0, align='center', alpha=1.0, color=cmap_out.to_rgba(x1))
small_ax_2.bar(np.arange(len(x2>0)), x2>0, align='center', alpha=0.8, color=cmap_in_x2.to_rgba(x2))
small_ax_2.bar(np.arange(len(x2<0)), x2<0, align='center', alpha=1.0, color=cmap_out.to_rgba(x2))

big_ax.set_xticks(tick_location)
big_ax.set_xticklabels(times, rotation=90)

big_ax.plot(np.zeros(len(x1)), color='black', linewidth=2)

plt.setp(small_ax_1.get_yticklabels(), visible=False)
plt.setp(small_ax_1.get_xticklabels(), visible=False)
plt.setp(small_ax_2.get_yticklabels(), visible=False)
plt.setp(small_ax_2.get_xticklabels(), visible=False)

plt.title('Bus Traffic', fontsize=24)
plt.show()

The subplot idea seems very good.

For the top graphs you might like to consider axhspan instead of bars. Bars tend to have the same width which might complicate your work. with axhspan you just define from where to where you want a bar to be drawn.

A simple example are the middle bars in this image, which indicate the daylight zones at the observer and the transmitter sites:

The code is something like this:

s.axhspan(y1, y2, xmin = 0., xmax = set, ec = 'black', fc = 'white', lw = 1)
s.axhspan(y1, y2, xmin = set, xmax = rise, linewidth=4, color='black', lw = 1)
s.axhspan(y1, y2, xmin = rise, xmax = 1., ec = 'black', fc = 'white', lw = 1)

where the xmin/xmax are the changes in day/night (in my case), but would be the opening/closing of the doors.

The plot preparation (in my plot) is like this:

s = plt.subplot(where, ylabel='Sun', xlim = (0., 24.), ylim = (0., 1.), \
                xticks = xticks, \
                yticks = [0.25, 0.75], yticklabels = [id, 'Obs'], \
                axisbg = '#00c000')

And, in the same plot, draws both bars, just specifying their heights (ymin and ymax)