In this toy example, I add Mario to a plot using add_artist. When I do that, I can't seem to clear the figure. Python throws RuntimeError: Can not put single artist in more than one figure when it tries to add mario to the second plot (02.png). Why is this happening? How can I avoid this error? I tried sending a copy of the AnnotationBbox to add_artist, following this approach, but it did not work.
import matplotlib.pyplot as plt
from matplotlib.offsetbox import OffsetImage, AnnotationBbox
vortexRF = plt.imread('./mario.png')
imagebox = OffsetImage(vortexRF, zoom=0.03)
for ii in range(3):
fig, ax = plt.subplots(2, 2)
plt.subplots_adjust(wspace=0.6, hspace=0.5)
for jj in range(2):
for kk in range(2):
ax[jj, kk].plot([0, 1], [0, 1], label='1')
ax[jj, kk].plot([0, 1], [0, 1], label='2', ls='--')
ax[1, 0].legend(loc='upper center', bbox_to_anchor=(.08, 2.85))
if True: # Switch to control if we add mario
ab = AnnotationBbox(imagebox, (0, 0), frameon=False)
cbar_ax = fig.add_axes([0.7, .92, 0.1, 0.1])
cbar_ax.add_artist(ab)
cbar_ax.axis('off')
plt.savefig('./%02d' % ii)
# attempt to clear figure
plt.clf()
plt.cla()
plt.close('all')
ab.remove()
If you are trying to make Mario run in a rush :) like shown below, I think you need to make a new 'imagebox' every time you add to the axis.
import matplotlib.pyplot as plt
from matplotlib.offsetbox import OffsetImage, AnnotationBbox
vortexRF = plt.imread('mario.png')
for ii in range(9):
fig, ax = plt.subplots(2, 2)
plt.subplots_adjust(wspace=0.6, hspace=0.5)
for jj in range(2):
for kk in range(2):
ax[jj, kk].plot([0, 1], [0, 1], label='1')
ax[jj, kk].plot([0, 1], [0, 1], label='2', ls='--')
ax[1, 0].legend(loc='upper center', bbox_to_anchor=(.08, 2.85))
if True: # Switch to control if we add mario
imagebox = OffsetImage(vortexRF, zoom=0.03)
ab = AnnotationBbox(imagebox, (0, 0), frameon=False)
cbar_ax = fig.add_axes([0.1+0.1*ii, .92, 0.1, 0.1])
cbar_ax.add_artist(ab)
cbar_ax.axis('off')
plt.savefig(str(ii)+'.png')
plt.show()
Related
I have a dataframe and I want to plot the seaborn heatmap:
import seaborn as sns
res = sns.heatmap(df, cmap='flare',xticklabels=1, yticklabels=1,linecolor='white',linewidths=0.5,
cbar=True,mask=df.isnull(),cbar_kws={'shrink': 0.6},vmin=vmin, vmax=vmax)
I have applied a mask for NaN cells. Now, I want to change the color of few cells by a customized color which is not in the colormap, for example blue, to show that those cells belong to another category.
My question is:
Is it possible to apply 2 masks or more with different colors to a seaborn heatmap or manually change the color of a cell to totally another color?
It is unclear how the blue squares are represented. The following solution supposes they are represented as ones in a second matrix. The first heatmap is drawn as before. Then the second heatmap uses a special colormap (in this case with one color, but also a full range is possible), masking out all places where nothing should be drawn.
Note that masks can be combined via the logical or (symbol: |).
from matplotlib import pyplot as plt
from matplotlib.colors import ListedColormap
import seaborn as sns
import pandas as pd
import numpy as np
N = 10
data = np.random.uniform(0, 45, size=(N, N))
for x, y in np.random.randint(0, N, 50).reshape(-1, 2):
data[x, y] = np.nan # fill in some nans at random places
df = pd.DataFrame(data)
up_triang = np.triu(np.ones_like(data)).astype(bool)
ax = sns.heatmap(df, cmap='flare', xticklabels=True, yticklabels=True, square=True,
linecolor='white', linewidths=0.5,
cbar=True, mask=df.isnull() | up_triang, cbar_kws={'shrink': 0.6, 'pad': 0}, vmin=0, vmax=45)
data_special = np.random.randint(0, 5, size=(N, N)) // 4
sns.heatmap(data_special, cmap=ListedColormap(['cornflowerblue']), linecolor='white', linewidths=0.5,
square=True, cbar=False, mask=(data_special != 1) | up_triang, ax=ax)
ax.plot([0, N, 0, 0], [0, N, N, 0], clip_on=False, color='black', lw=2)
ax.tick_params(left=False, bottom=False)
plt.show()
An alternative approach, when there is only one color for the special cells, is to use an "under" color for the colormap, and give these cells negative values. An additional benefit is that the color can be shown in the colorbar. Here is some sample code:
N = 10
data = np.random.uniform(0, 45, size=(N, N))
for x, y in np.random.randint(0, N, 50).reshape(-1, 2):
data[x, y] = np.nan
data_special = np.random.randint(0, 5, size=(N, N)) // 4
data[data_special == 1] = -1
df = pd.DataFrame(data)
up_triang = np.triu(np.ones_like(data)).astype(bool)
cmap = sns.color_palette('mako', as_cmap=True).copy()
cmap.set_under('crimson ')
ax = sns.heatmap(df, cmap=cmap, xticklabels=True, yticklabels=True, square=True,
linecolor='white', linewidths=0.5, cbar=True, mask=df.isnull() | up_triang,
cbar_kws={'shrink': 0.6, 'pad': 0, 'extend': 'min', 'extendrect': True}, vmin=0, vmax=45)
ax.plot([0, N, 0, 0], [0, N, N, 0], clip_on=False, color='black', lw=2)
ax.tick_params(left=False, bottom=False)
plt.show()
Suppose something comes up in my plot that mandates that I change the height ratio between two subplots that I've generated within my plot. I've tried changing GridSpec's height ratio to no avail.
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec
fig = plt.figure()
gs = GridSpec(2, 1, height_ratios=[2, 1])
ax1 = fig.add_subplot(gs[0])
ax1 = fig.axes[0]
ax2 = fig.add_subplot(gs[1])
ax2 = fig.axes[1]
ax1.plot([0, 1], [0, 1])
ax2.plot([0, 1], [1, 0])
gs.height_ratios = [2, 5]
The last line has no effect on the plot ratio.
In my actual code, it is not feasible without major reworking to set the height_ratios to 2:5 ahead of time.
How do I get this to update like I want?
The axes of relevant subplots can be manipulated and adjusted to get new height ratios.
import matplotlib.pyplot as plt
from matplotlib.gridspec import GridSpec
fig = plt.figure()
gs = GridSpec(2, 1, height_ratios=[2, 1]) #nrows, ncols
ax1 = fig.add_subplot(gs[0])
ax1 = fig.axes[0]
ax2 = fig.add_subplot(gs[1])
ax2 = fig.axes[1]
ax1.plot([0, 1], [0, 1])
ax2.plot([0, 1], [1, 0])
# new height ratio: 2:5 is required for the 2 subplots
rw, rh = 2, 5
# get dimensions of the 2 axes
box1 = ax1.get_position()
box2 = ax2.get_position()
# current dimensions
w1,h1 = box1.x1-box1.x0, box1.y1-box1.y0
w2,h2 = box2.x1-box2.x0, box2.y1-box2.y0
top1 = box1.y0+h1
#top2 = box2.y0+h2
full_h = h1+h2 #total height
# compute new heights for each axes
new_h1 = full_h*rw/(rw + rh)
new_h2 = full_h*rh/(rw + rh)
#btm1,btm2 = box1.y0, box2.y0
new_bottom1 = top1-new_h1
# finally, set new location/dimensions of the axes
ax1.set_position([box1.x0, new_bottom1, w1, new_h1])
ax2.set_position([box2.x0, box2.y0, w2, new_h2])
plt.show()
The output for ratio: (2, 5):
The output for (2, 10):
Consider the following code:
import matplotlib.pyplot as plt
import numpy as np
from pylab import *
graph_data = [[0, 1, 2, 3], [5, 8, 7, 9]]
x = range(len(graph_data[0]))
y = graph_data[1]
fig, ax = plt.subplots()
alpha = 0.5
plt.plot(x, y, '-o',markersize=3, color=[1., alpha, alpha], markeredgewidth=0.0)
ax.fill_between(x, 0, y, facecolor=[1., alpha, alpha], interpolate=False)
plt.show()
filename = 'test1.pdf'
fig.savefig(filename, bbox_inches='tight')
It works fine. However, when zoomed in the generated PDF, I can see two thin gray/black boundaries that separate the line:
I can see this when viewing in both Edge and Chrome. My question is, how can I get rid of the boundaries?
UPDATE I forgot to mention, I was using Sage to generate the graph. Now it seems a problem specific to Sage (and not to Python in general). This time I used native Python, and got correct result.
I could not reproduce it but maybe you can try to not plot the line.
import matplotlib.pyplot as plt
import numpy as np
from pylab import *
graph_data = [[0, 1, 2, 3], [5, 8, 7, 9]]
x = range(len(graph_data[0]))
y = graph_data[1]
fig, ax = plt.subplots()
alpha = 0.5
plt.plot(x, y, 'o',markersize=3, color=[1., alpha, alpha])
ax.fill_between(x, 0, y, facecolor=[1., alpha, alpha], interpolate=False)
plt.show()
filename = 'test1.pdf'
fig.savefig(filename, bbox_inches='tight')
I am trying to make a bunch of polar view plots using the same colorbar. However, the colorbars differ after setting the plotting limits. In the code snippet below, I randomly created 5 maps but plotted in a fixed range, but the output figures are still different in colorbar.
from numpy import linspace, pi, ndarray, random
import matplotlib
matplotlib.use('Agg')
from matplotlib.pyplot import figure
lon = linspace(start=0, stop=2*pi, num=100)
colat = linspace(start=0, stop=9, num=10)
emission = ndarray(shape=(10, 100, 5), dtype=float)
for t in range(5):
emission[:, :, t] = random.rand(10, 100)
fig = figure(num='emission', figsize=(15, 15))
em_pos = [0.05, 0.1, 0.8, 0.8]
emc_pos = [0.9, 0.1, 0.05, 0.8]
for t in range(5):
fig.clear()
ax = fig.add_subplot(121, polar=True, position=em_pos)
axcont = ax.contourf(lon, colat, emission[:, :, t], vmin=0, vmax=2)
axc = fig.add_subplot(122, position=emc_pos)
fig.colorbar(mappable=axcont, cax=axc)
fig.savefig(fname='emission{0:d}.png'.format(t), format='png')
The problem seems to be solved. It is not a problem of colorbar, but a problem of contourf. When I replaced
ax.contourf(lon, colat, emission[:, :, t], vmin=0, vmax=2)
with
ax.pcolormesh(lon, colat, emission[:, :, t], vmin=0, vmax=2)
Then the colorbar shows the proper range. Indeed it is not a full solution, pcolormesh differs in some aspects from contourf, but it meets my needs.
I'm plotting a series of data points with x and y error but do NOT want the errorbars to be included in the legend (only the marker). Is there a way to do so?
Example:
import matplotlib.pyplot as plt
import numpy as np
subs=['one','two','three']
x=[1,2,3]
y=[1,2,3]
yerr=[2,3,1]
xerr=[0.5,1,1]
fig,(ax1)=plt.subplots(1,1)
for i in np.arange(len(x)):
ax1.errorbar(x[i],y[i],yerr=yerr[i],xerr=xerr[i],label=subs[i],ecolor='black',marker='o',ls='')
ax1.legend(loc='upper left', numpoints=1)
fig.savefig('test.pdf', bbox_inches=0)
You can modify the legend handler. See the legend guide of matplotlib.
Adapting your example, this could read:
import matplotlib.pyplot as plt
import numpy as np
subs=['one','two','three']
x=[1,2,3]
y=[1,2,3]
yerr=[2,3,1]
xerr=[0.5,1,1]
fig,(ax1)=plt.subplots(1,1)
for i in np.arange(len(x)):
ax1.errorbar(x[i],y[i],yerr=yerr[i],xerr=xerr[i],label=subs[i],ecolor='black',marker='o',ls='')
# get handles
handles, labels = ax1.get_legend_handles_labels()
# remove the errorbars
handles = [h[0] for h in handles]
# use them in the legend
ax1.legend(handles, labels, loc='upper left',numpoints=1)
plt.show()
This produces
Here is an ugly patch:
pp = []
colors = ['r', 'b', 'g']
for i, (y, yerr) in enumerate(zip(ys, yerrs)):
p = plt.plot(x, y, '-', color='%s' % colors[i])
pp.append(p[0])
plt.errorbar(x, y, yerr, color='%s' % colors[i])
plt.legend(pp, labels, numpoints=1)
Here is a figure for example:
The accepted solution works in simple cases but not in general. In particular, it did not work in my own more complex situation.
I found a more robust solution, which tests for ErrorbarContainer, which did work for me. It was proposed by Stuart W D Grieve and I copy it here for completeness
import matplotlib.pyplot as plt
from matplotlib import container
label = ['one', 'two', 'three']
color = ['red', 'blue', 'green']
x = [1, 2, 3]
y = [1, 2, 3]
yerr = [2, 3, 1]
xerr = [0.5, 1, 1]
fig, (ax1) = plt.subplots(1, 1)
for i in range(len(x)):
ax1.errorbar(x[i], y[i], yerr=yerr[i], xerr=xerr[i], label=label[i], color=color[i], ecolor='black', marker='o', ls='')
handles, labels = ax1.get_legend_handles_labels()
handles = [h[0] if isinstance(h, container.ErrorbarContainer) else h for h in handles]
ax1.legend(handles, labels)
plt.show()
It produces the following plot (on Matplotlib 3.1)
I works for me if I set the label argument as a None type.
plt.errorbar(x, y, yerr, label=None)