Set the background color on a plot in specific alternate intervals of dates or other types of values

Question

I want to do a plot similar to yahoo finance charts where the background color is greyed out in alternate intervals according to the axis ticks date marks. Following an answer from a similar problem I get an image like this:

Using this code:

n = 1000
xs = np.random.randn(n).cumsum()

plt.plot(xs)
plt.autoscale(enable=True, axis='both', tight=True)

for i in range(0, len(y_series), 400):
  plt.axvspan(i, i+100, facecolor='grey', alpha=0.5)

However, the issue with this code is that we use the data input as a reference for determining the greyed out area. Instead, I want the greyed out area to be determined by the visible ticks on the x-axis or y-axis, decoupled from the input. I do not want to have to use the locator functions, because this also defeated the purpose of 'automatically' greying out the background according to the visible ticks values. Additionally, we used integers in the x-axis, but ideally, this should work for dates, floats, and others.

Here is an example using dates, without the greyed out areas:

Produced with this code and without autoscale:

n = 700
x_series = pd.date_range(start='2017-01-01', periods=n, freq='D')
y_series = np.random.randn(n).cumsum()

fig, ax = plt.subplots()
ax.plot(x_series, y_series)
plt.gcf().autofmt_xdate()

PS: I tried reading the sticks list, but that list does not reflect exactly the visible values if autoscale if turned off.

locs, labels = plt.xticks()
print(locs)

[-200. 0. 200. 400. 600. 800. 1000. 1200.]

Answer 1

I think this problem is a bit fiddly, where it's tedious to make sure all cases are covered. It's true that xticks sometimes returns values to the left and right of the xlim, but that's only the start. What happens if the data extends beyond the rightmost xtick, or starts before the leftmost xtick, etc?

For example, in many of the cases below, I want to start (or stop) the bands at xmin or xmax, because if I don't and the indexing skips the bands after the ticks start (or stop), there would be a long section that was unbanded and it wouldn't look right.

So in playing around with a few different corner cases, I have landed on this as covering (at least) the ones that I tried:

import numpy as np
import matplotlib.pyplot as plt

x = np.linspace(100, 11000, 7500)
y = x * np.sin(0.001*x) ** 2

def alt_bands(x):
    locs, labels = plt.xticks()
    x_left, x_right = plt.xlim()
    for i, loc in enumerate(locs):
        if i%2 == 1 and i<len(locs)-1 and loc<x[-1] and (loc>x_left or x[0]>x_left):
            L = max(x_left, x[0], loc)
            R = min(x_right, x[-1], locs[i+1])
            if x[0] <= L and R>L:
                plt.axvspan(L, R, facecolor='grey', alpha=0.5)

plt.plot(x, y)
alt_bands()

And here are some example plots:

Honestly, this is not the SO answer I'm the most proud of. I didn't carefully think through the logic, but instead progressively added conditions to deal with each new corner case I tried but in a way that didn't bump into the previous case. Please feel free to clean it up if you want to think it through. Or is there a way that's intrinsically clean?

Answer 2

As the comments seem to be too complicated to explain everything, here is some example code, including subplots, autoscale, autofmt_xdate and resetting the xlims.

autoscale moves the xlims, so it should be called before alt_bands gets and resets these xlims.

When working with subplots, most functions should be the axes version instead of the plt versions. So, ax.get_ticks() instead of plt.ticks() and ax.axvspan instead of plt.axvspan. autofmt_xdate changes the complete figure (rotates and realigns the dates on x-axes, and removes dates on x-axes except the ones of the plots at the bottom). fig.autofmt_xdate() should be called after creating the plot (after ax.plot) and after operations that might change tick positions.

import numpy as np
import pandas as pd
from matplotlib import pyplot as plt

def alt_bands(ax=None):
    ax = ax or plt.gca()
    x_left, x_right = ax.get_xlim()
    locs = ax.get_xticks()
    for loc1, loc2 in zip(locs[::2], np.concatenate((locs, [x_right]))[1::2]):
        ax.axvspan(loc1, loc2, facecolor='black', alpha=0.2)
    ax.set_xlim(x_left, x_right)

n = 700
x_series = pd.date_range(start='2017-01-01', periods=n, freq='D')
y_series = np.random.normal(.01, 1, n).cumsum()

fig, axes = plt.subplots(ncols=2)
axes[0].plot(x_series, y_series)
axes[0].autoscale(enable=True, axis='both', tight=True)
alt_bands(axes[0])

axes[1].plot(x_series[200:400], y_series[200:400])
axes[1].autoscale(enable=True, axis='both', tight=True)
alt_bands(axes[1])

fig.autofmt_xdate()
plt.show()

Answer 3

check that code could help you !

here axes([0.01, 0.01, 0.98, 0.90], facecolor="white", frameon=True) in facecolor you can change the background, also with hex format '#F0F0F0' gray color

from matplotlib.pyplot import *
import subprocess
import sys
import re

# Selection of features following "Writing mathematical expressions" tutorial
mathtext_titles = {
    0: "Header demo",
    1: "Subscripts and superscripts",
    2: "Fractions, binomials and stacked numbers",
    3: "Radicals",
    4: "Fonts",
    5: "Accents",
    6: "Greek, Hebrew",
    7: "Delimiters, functions and Symbols"}
n_lines = len(mathtext_titles)

# Randomly picked examples
mathext_demos = {
    0: r"$W^{3\beta}_{\delta_1 \rho_1 \sigma_2} = "
       r"U^{3\beta}_{\delta_1 \rho_1} + \frac{1}{8 \pi 2} "
       r"\int^{\alpha_2}_{\alpha_2} d \alpha^\prime_2 \left[\frac{ "
       r"U^{2\beta}_{\delta_1 \rho_1} - \alpha^\prime_2U^{1\beta}_"
       r"{\rho_1 \sigma_2} }{U^{0\beta}_{\rho_1 \sigma_2}}\right]$",

    1: r"$\alpha_i > \beta_i,\ "
       r"\alpha_{i+1}^j = {\rm sin}(2\pi f_j t_i) e^{-5 t_i/\tau},\ "
       r"\ldots$",

    2: r"$\frac{3}{4},\ \binom{3}{4},\ \genfrac{}{}{0}{}{3}{4},\ "
       r"\left(\frac{5 - \frac{1}{x}}{4}\right),\ \ldots$",

    3: r"$\sqrt{2},\ \sqrt[3]{x},\ \ldots$",

    4: r"$\mathrm{Roman}\ , \ \mathit{Italic}\ , \ \mathtt{Typewriter} \ "
       r"\mathrm{or}\ \mathcal{CALLIGRAPHY}$",

    5: r"$\acute a,\ \bar a,\ \breve a,\ \dot a,\ \ddot a, \ \grave a, \ "
       r"\hat a,\ \tilde a,\ \vec a,\ \widehat{xyz},\ \widetilde{xyz},\ "
       r"\ldots$",

    6: r"$\alpha,\ \beta,\ \chi,\ \delta,\ \lambda,\ \mu,\ "
       r"\Delta,\ \Gamma,\ \Omega,\ \Phi,\ \Pi,\ \Upsilon,\ \nabla,\ "
       r"\aleph,\ \beth,\ \daleth,\ \gimel,\ \ldots$",

    7: r"$\coprod,\ \int,\ \oint,\ \prod,\ \sum,\ "
       r"\log,\ \sin,\ \approx,\ \oplus,\ \star,\ \varpropto,\ "
       r"\infty,\ \partial,\ \Re,\ \leftrightsquigarrow, \ \ldots$"}


def doall():
    # Colors used in mpl online documentation.
    mpl_blue_rvb = (191. / 255., 209. / 256., 212. / 255.)
    mpl_orange_rvb = (202. / 255., 121. / 256., 0. / 255.)
    mpl_grey_rvb = (51. / 255., 51. / 255., 51. / 255.)

    # Creating figure and axis.
    figure(figsize=(6, 7))
    axes([0.01, 0.01, 0.98, 0.90], facecolor="white", frameon=True)
    gca().set_xlim(0., 1.)
    gca().set_ylim(0., 1.)
    gca().set_title("Matplotlib's math rendering engine",
                    color=mpl_grey_rvb, fontsize=14, weight='bold')
    gca().set_xticklabels("", visible=False)
    gca().set_yticklabels("", visible=False)

    # Gap between lines in axes coords
    line_axesfrac = (1. / n_lines)

    # Plotting header demonstration formula
    full_demo = mathext_demos[0]
    annotate(full_demo,
             xy=(0.5, 1. - 0.59 * line_axesfrac),
             color=mpl_orange_rvb, ha='center', fontsize=20)

    # Plotting features demonstration formulae
    for i_line in range(1, n_lines):
        baseline = 1 - i_line * line_axesfrac
        baseline_next = baseline - line_axesfrac
        toptitle = mathtext_titles[i_line] + ":"
        fill_color = ['white', mpl_blue_rvb][i_line % 2]
        fill_between([0., 1.], [baseline, baseline],
                     [baseline_next, baseline_next],
                     color=fill_color, alpha=0.5)
        annotate(toptitle,
                 xy=(0.07, baseline - 0.3 * line_axesfrac),
                 color=mpl_grey_rvb, weight='bold')
        demo = mathext_demos[i_line]
        annotate(demo,
                 xy=(0.05, baseline - 0.75 * line_axesfrac),
                 color=mpl_grey_rvb, fontsize=16)

    for i1 in range(n_lines):
        s1 = mathext_demos[i1]
        print(i1, s1)
    show()


if __name__ == '__main__':
    if '--latex' in sys.argv:
        # Run: python mathtext_examples.py --latex
        # Need amsmath and amssymb packages.
        fd = open("mathtext_examples.ltx", "w")
        fd.write("\\documentclass{article}\n")
        fd.write("\\usepackage{amsmath, amssymb}\n")
        fd.write("\\begin{document}\n")
        fd.write("\\begin{enumerate}\n")

        for i in range(n_lines):
            s = mathext_demos[i]
            s = re.sub(r"(?<!\\)\$", "$$", s)
            fd.write("\\item %s\n" % s)

        fd.write("\\end{enumerate}\n")
        fd.write("\\end{document}\n")
        fd.close()

        subprocess.call(["pdflatex", "mathtext_examples.ltx"])
    else:
        doall()