如何使用matplotlib和python更改多线图上的颜色？

Question

I am using a dictionary to plot lines with python and matplotlib i can't figure out why the colors of my lines are not changing

from datetime import datetime
import matplotlib.pyplot as plt


dico =  {'A01': [(u'11/10/12-08:00:01', 2.0), (u'11/10/12-08:10:00', 10.0), \
                 (u'11/10/12-08:20:01', 5.0), (u'11/10/12-08:30:01', 15.0), \
                 (u'11/10/12-08:40:00', 7.0), (u'11/10/12-08:50:01', 45.0)],
         'A02': [(u'11/10/12-08:00:01', 10.0), (u'11/10/12-08:10:00', 12.0), \
                 (u'11/10/12-08:20:01', 15.0), (u'11/10/12-08:30:01', 10.0), \
                 (u'11/10/12-08:40:00', 17.0), (u'11/10/12-08:50:01', 14.0)]}

x = []
y = []
lstPlot = []
plt.gca().set_color_cycle(["b", "g", "r", "c", "m", "y", "k"])
for key, values in dico.iteritems():
    for i in  sorted(values):
        # date sting to date obj
        dateObj = datetime.strptime(i[0], "%d/%m/%y-%H:%M:%S")
        line = dateObj, i[1]
        lstPlot.append(line)
    for i in sorted(lstPlot):
        x.append(i[0])
        y.append(i[1])
    plt.plot(x, y, label=key)



# plotting

plt.legend(loc='upper right')
plt.xlabel('Dates')
plt.ylabel("titre")
plt.title("Modbus")
plt.show()

Note that I have different colors in the legend but not in the plot.

Answer 1

They are changing, but you're overplotting one with the other. These lines

x = []
y = []
lstPlot = []

need to be inside the loop. Otherwise lstPlot will just grow. For example, adding print lstPlot inside the loop gives:

[(datetime.datetime(2012, 10, 11, 8, 0, 1), 10.0), (datetime.datetime(2012, 10, 11, 8, 10), 12.0), (datetime.datetime(2012, 10, 11, 8, 20, 1), 15.0), (datetime.datetime(2012, 10, 11, 8, 30, 1), 10.0), (datetime.datetime(2012, 10, 11, 8, 40), 17.0), (datetime.datetime(2012, 10, 11, 8, 50, 1), 14.0)]
[(datetime.datetime(2012, 10, 11, 8, 0, 1), 10.0), (datetime.datetime(2012, 10, 11, 8, 10), 12.0), (datetime.datetime(2012, 10, 11, 8, 20, 1), 15.0), (datetime.datetime(2012, 10, 11, 8, 30, 1), 10.0), (datetime.datetime(2012, 10, 11, 8, 40), 17.0), (datetime.datetime(2012, 10, 11, 8, 50, 1), 14.0), (datetime.datetime(2012, 10, 11, 8, 0, 1), 2.0), (datetime.datetime(2012, 10, 11, 8, 10), 10.0), (datetime.datetime(2012, 10, 11, 8, 20, 1), 5.0), (datetime.datetime(2012, 10, 11, 8, 30, 1), 15.0), (datetime.datetime(2012, 10, 11, 8, 40), 7.0), (datetime.datetime(2012, 10, 11, 8, 50, 1), 45.0)]

(You'll probably have to scroll over to see that the second list is a lot longer than the first, but you should notice that the first value is the same in both.)

So you can either clear the lists inside, or maybe you can simplify it a bit:

for key, values in dico.iteritems():
    points = [(datetime.strptime(i[0], "%d/%m/%y-%H:%M:%S"), i[1]) for i in values]
    points.sort()
    x, y = zip(*points)
    plt.plot(x, y, label=key)

That code, with the addition of @bmu's suggestion of

plt.gcf().autofmt_xdate()

to automatically make the x-axis look nice, produces

[Alternatively, you might want to use get_xticklabels() and methods like set_rotation for finer control.]

Answer 2

Try :

colors = ["b", "g", "r", "c", "m", "y", "k"]
for (key, values), c in zip(dico.iteritems(), colors):
    ...
    plt.plot(x, y, c, label=key)