Here is an example that reproduces my problem:
import matplotlib.pyplot as plt
import numpy as np
data1,data2,data3,data4 = np.random.random(100),np.random.random(100),np.random.random(100),np.random.random(100)
fig,ax = plt.subplots()
ax.plot(data1)
ax.plot(data2)
ax.plot(data3)
ax2 = ax.twinx()
ax2.plot(data4)
plt.grid('on')
ax.legend(['1','2','3'], loc='center')
ax2.legend(['4'], loc=1)
How can I get the legend in the center to plot on top of the lines?
To get exactly what you have asked for, try the following. Note I have modified your code to define the labels when you generate the plot and also the colors so you don't get a repeated blue line.
import matplotlib.pyplot as plt
import numpy as np
data1,data2,data3,data4 = (np.random.random(100),
np.random.random(100),
np.random.random(100),
np.random.random(100))
fig,ax = plt.subplots()
ax.plot(data1, label="1", color="k")
ax.plot(data2, label="2", color="r")
ax.plot(data3, label="3", color="g")
ax2 = ax.twinx()
ax2.plot(data4, label="4", color="b")
# First get the handles and labels from the axes
handles1, labels1 = ax.get_legend_handles_labels()
handles2, labels2 = ax2.get_legend_handles_labels()
# Add the first legend to the second axis so it displaysys 'on top'
first_legend = plt.legend(handles1, labels1, loc='center')
ax2.add_artist(first_legend)
# Add the second legend as usual
ax2.legend(handles2, labels2)
plt.show()
Now I will add that it would be clearer if you just use a single legend adding all the lines to that. This is described in this SO post and in the code above can easily be achieved with
ax2.legend(handles1+handles2, labels1+labels2)
But obviously you may have your own reasons for wanting two legends.
Related
I want to create a heatmap with seaborn, similar to this (with the following code):
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
import numpy as np
# Create data
df = pd.DataFrame(np.random.random((5,5)), columns=["a","b","c","d","e"])
# Default heatmap
ax = sns.heatmap(df)
plt.show()
I'd also like to add a new variable (lets say new_var = pd.DataFrame(np.random.random((5,1)), columns=["new variable"])), such as that the values (and possibly the spine and ticks as well) of the y-axis are colored according to the new variable and a second color bar plotted in the same plot to represent the colors of the y-axis values. How can I do that?
This uses the new values to color the y-ticks and the y-tick labels and adds the associated colorbar.
import matplotlib.pyplot as plt
import matplotlib
import seaborn as sns
import pandas as pd
import numpy as np
# Create data
df = pd.DataFrame(np.random.random((5,5)), columns=["a","b","c","d","e"])
# Default heatmap
ax = sns.heatmap(df)
new_var = pd.DataFrame(np.random.random((5,1)), columns=["new variable"])
# Create the colorbar for y-ticks and labels
norm = plt.Normalize(new_var.min(), new_var.max())
cmap = matplotlib.cm.get_cmap('turbo')
yticks_locations = ax.get_yticks()
yticks_labels = df.index.values
#hide original ticks
ax.tick_params(axis='y', left=False)
ax.set_yticklabels([])
for var, ytick_loc, ytick_label in zip(new_var.values, yticks_locations, yticks_labels):
color = cmap(norm(float(var)))
ax.annotate(ytick_label, xy=(1, ytick_loc), xycoords='data', xytext=(-0.4, ytick_loc),
arrowprops=dict(arrowstyle="-", color=color, lw=1), zorder=0, rotation=90, color=color)
# Add colorbar for y-tick colors
sm = plt.cm.ScalarMappable(cmap=cmap, norm=norm)
cb = ax.figure.colorbar(sm)
# Match the seaborn style
cb.outline.set_visible(False)
I found your problem interesting, and inspired by the unanswered comment above:
How do you change the second colorbar position? For example, one on top the other on bottom sides. - Py-ser
I decided to spend a while doing some tests. After a little digging i find that cbar_kws={"orientation": "horizontal"} is the argument for sns.heatmap that makes the colorbars horizontal.
Borrowing the code from the solution and making some changes, you can format your plot the way you want as in:
import matplotlib.pyplot as plt
import matplotlib
import seaborn as sns
import pandas as pd
import numpy as np
# Create data
df = pd.DataFrame(np.random.random((5,5)), columns=["a","b","c","d","e"])
# Default heatmap
ax = sns.heatmap(df, cbar_kws={"orientation": "horizontal"}, square = False, annot = True)
new_var = pd.DataFrame(np.random.random((5,1)), columns=["new variable"])
# Create the colorbar for y-ticks and labels
norm = plt.Normalize(new_var.min(), new_var.max())
cmap = matplotlib.cm.get_cmap('turbo')
yticks_locations = ax.get_yticks()
yticks_labels = df.index.values
#hide original ticks
ax.tick_params(axis='y', left=False)
ax.set_yticklabels([])
for var, ytick_loc, ytick_label in zip(new_var.values, yticks_locations, yticks_labels):
color = cmap(norm(float(var)))
ax.annotate(ytick_label, xy=(1, ytick_loc), xycoords='data', xytext=(-0.4, ytick_loc),
arrowprops=dict(arrowstyle="-", color=color, lw=1), zorder=0, rotation=90, color=color)
# Add colorbar for y-tick colors
sm = plt.cm.ScalarMappable(cmap=cmap, norm=norm)
cb = ax.figure.colorbar(sm)
# Match the seaborn style
cb.outline.set_visible(False)
Also, you will notice that I listed the values related to each cell in the heatmap, but just out of curiosity to make it clearer to check that everything was working as expected.
I'm still not very happy with the shape/size of the horizontal colorbar, but I'll keep testing and update any progress by editing this answer!
==========================================
EDIT
just to keep track of the updates, first i tried to change just some parameters of seaborn's heatmap function but wouldn't consider this a major improvement on the task... by adding
ax = sns.heatmap(df, cbar_kws = dict(use_gridspec=True, location="top", shrink =0.6), square = True, annot = True)
I end up with:
I did get to separate the colormap using the matplotlib subplot routine and honestly i believe this is the right way given the parameter control that is possible to get here, by:
# Define two rows for subplots
fig, (cax, ax) = plt.subplots(nrows=2, figsize=(5,5.025), gridspec_kw={"height_ratios":[0.025, 1]})
# Default heatmap
ax = sns.heatmap(df, cbar=False, annot = True)
# colorbar
fig.colorbar(ax.get_children()[0], cax=cax, orientation="horizontal")
plt.show()
I obtained:
Which is still not the prettiest graph I've ever made, but now the position and size of the heatmap can be edited normally within the plt.subplots subroutines that give absolute control over these parameters.
I would like to make a paired histogram like the one shown here using the seaborn distplot.
This kind of plot can also be referred to as the back-to-back histogram shown here, or a bihistogram inverted/mirrored along the x-axis as discussed here.
Here is my code:
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
green = np.random.normal(20,10,1000)
blue = np.random.poisson(60,1000)
fig, ax = plt.subplots(figsize=(8,6))
sns.distplot(blue, hist=True, kde=True, hist_kws={'edgecolor':'black'}, kde_kws={'linewidth':2}, bins=10, color='blue')
sns.distplot(green, hist=True, kde=True, hist_kws={'edgecolor':'black'}, kde_kws={'linewidth':2}, bins=10, color='green')
ax.set_xticks(np.arange(-20,121,20))
ax.set_yticks(np.arange(0.0,0.07,0.01))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
plt.show()
Here is the output:
When I use the method discussed here (plt.barh), I get the bar plot shown just below, which is not what I am looking for.
Or maybe I haven't understood the workaround well enough...
A simple/short implementation of python-seaborn-distplot similar to these kinds of plots would be perfect. I edited the figure of my first plot above to show the kind of plot I hope to achieve (though y-axis not upside down):
Any leads would be greatly appreciated.
You could use two subplots and invert the y-axis of the lower one and plot with the same bins.
df = pd.DataFrame({'a': np.random.normal(0,5,1000), 'b': np.random.normal(20,5,1000)})
fig =plt.figure(figsize=(5,5))
ax = fig.add_subplot(211)
ax2 = fig.add_subplot(212)
bins = np.arange(-20,40)
ax.hist(df['a'], bins=bins)
ax2.hist(df['b'],color='orange', bins=bins)
ax2.invert_yaxis()
edit:
improvements suggested by #mwaskom
fig, axes = plt.subplots(nrows=2, ncols=1, sharex=True, figsize=(5,5))
bins = np.arange(-20,40)
for ax, column, color, invert in zip(axes.ravel(), df.columns, ['teal', 'orange'], [False,True]):
ax.hist(df[column], bins=bins, color=color)
if invert:
ax.invert_yaxis()
plt.subplots_adjust(hspace=0)
Here is a possible approach using seaborn's displots.
Seaborn doesn't return the created graphical elements, but the ax can be interrogated. To make sure the ax only contains the elements you want upside down, those elements can be drawn first. Then, all the patches (the rectangular bars) and the lines (the curve for the kde) can be given their height in negative. Optionally the x-axis can be set at y == 0 using ax.spines['bottom'].set_position('zero').
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
green = np.random.normal(20, 10, 1000)
blue = np.random.poisson(60, 1000)
fig, ax = plt.subplots(figsize=(8, 6))
sns.distplot(green, hist=True, kde=True, hist_kws={'edgecolor': 'black'}, kde_kws={'linewidth': 2}, bins=10,
color='green')
for p in ax.patches: # turn the histogram upside down
p.set_height(-p.get_height())
for l in ax.lines: # turn the kde curve upside down
l.set_ydata(-l.get_ydata())
sns.distplot(blue, hist=True, kde=True, hist_kws={'edgecolor': 'black'}, kde_kws={'linewidth': 2}, bins=10,
color='blue')
ax.set_xticks(np.arange(-20, 121, 20))
ax.set_yticks(np.arange(0.0, 0.07, 0.01))
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
pos_ticks = np.array([t for t in ax.get_yticks() if t > 0])
ticks = np.concatenate([-pos_ticks[::-1], [0], pos_ticks])
ax.set_yticks(ticks)
ax.set_yticklabels([f'{abs(t):.2f}' for t in ticks])
ax.spines['bottom'].set_position('zero')
plt.show()
Here is my (incomplete, I have note added the data itself) code, which produces a somewhat confusing plot, where one line is covered by the grid but the other not.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import pylab
sns.set_context("poster",font_scale=fs)
sns.set_style("darkgrid") # No grid lines
# sns.set_style({'legend.frameon': 'True'})
sns.set_style({'xtick.major.size':'0.0'})
c1,c2 = sns.color_palette("hls",2)#sns.color_palette("colorblind", 2)
a = sns.color_palette("BuGn_r")
# runs_plot = pd.DataFrame(runs.values+8.5)
# Plot just first state trajectory
fig, ax1 = plt.subplots(1,sharey=True, sharex=True, figsize=(30,8))
ax1.plot((ground.values+6),label='Ground Truth',color=c1)
ax1.set_xlabel('Time [$s$]')
ax1.set_ylim(0,10)
ax1.set_ylabel('State [$\#$]')
for tl in ax1.get_yticklabels():
tl.set_color(c1)
ax2 = ax1.twinx()
ax2.plot(0.4*signal_syn.values+1,color=c2,label='Emission Signal')
ax2.set_ylabel('Observations')
ax2.set_ylim(0,10)
# ax2.set_axisbelow(True)
for tl in ax2.get_yticklabels():
tl.set_color(c2)
# ask matplotlib for the plotted objects and their labels
lines, labels = ax1.get_legend_handles_labels()
lines2, labels2 = ax2.get_legend_handles_labels()
ax2.legend(lines + lines2, labels + labels2,ncol=5,loc='upper center', bbox_to_anchor=(0.5, -0.2))
plt.show()
which produces
now and you can probably see, that for the "Ground Truth" the line is covered by the 'darkgrid' option of the seaborn (which produces a white grid as seen above). Now for some reason the grid is not above the emission signal but only the ground truth.
Any ideas for why this might be?
So this is what I ended up doing, it is probably more of a hack than an actual solution, but it works. I just moved the plotting elements so that they're all plotted above the grid.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import pylab
sns.set_context("poster",font_scale=fs)
sns.set_style("darkgrid") # No grid lines
# sns.set_style({'legend.frameon': 'True'})
sns.set_style({'xtick.major.size':'0.0'})
c1,c2 = sns.color_palette("hls",2)#sns.color_palette("colorblind", 2)
a = sns.color_palette("BuGn_r")
# runs_plot = pd.DataFrame(runs.values+8.5)
# Plot just first state trajectory
fig, ax1 = plt.subplots(1,sharey=True, sharex=True, figsize=(30,8))
ax1.set_xlabel('Time [$s$]')
ax1.set_ylim(0,10)
ax1.set_ylabel('State [$\#$]')
for tl in ax1.get_yticklabels():
tl.set_color(c1)
ax2 = ax1.twinx()
ax2.plot((ground.values+6),label='Ground Truth',color=c1)
ax2.plot(0.4*signal_syn.values+1,color=c2,label='Emission Signal')
ax2.set_ylabel('Observations')
ax2.set_ylim(0,10)
# ax2.set_axisbelow(True)
for tl in ax2.get_yticklabels():
tl.set_color(c2)
# ask matplotlib for the plotted objects and their labels
lines, labels = ax1.get_legend_handles_labels()
lines2, labels2 = ax2.get_legend_handles_labels()
ax2.legend(lines + lines2, labels + labels2,ncol=5,loc='upper center', bbox_to_anchor=(0.5, -0.2))
plt.show()
Seems like the answer is in this question:
Matplotlib: draw grid lines behind other graph elements
And it is basically: Axis.set_axisbelow(True)
I have an issue where adding tick labels interferes with my given padding preference between subplots. What I want, is a tight_layout with no padding at all in between, but with some custom ticks along the x-axis. This snippet and resulting figures shows the issue:
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
fig_names = ['fig1']
gs = gridspec.GridSpec(1, len(fig_names))
gs.update(hspace=0.0)
figs = dict()
for fig_name in fig_names:
figs[fig_name] = plt.figure(figsize=(3*len(fig_names),6))
for i in range(0,len(fig_names)):
ax = figs[fig_name].add_subplot(gs[i])
ax.plot([0,1],[0,1], 'r-')
if i != 0:
ax.set_yticks(list())
ax.set_yticklabels(list())
ax.set_xticks(list())
ax.set_xticklabels(list())
for name,fig in figs.items():
fig.text(0.5, 0.03, 'Common xlabel', ha='center', va='center')
gs.tight_layout(fig, h_pad=0.0, w_pad=0.0)
ax = fig.add_subplot(gs[len(fig_names)-1])
ax.legend(('Some plot'), loc=2)
plt.show()
By changing the corresponding lines into:
ax.set_xticks([0.5,1.0])
ax.set_xticklabels(['0.5','1.0'])
...unwanted padding is added to the graphs.
How can I customize the tick text so that the graph plots has no padding, regardless of what tick text I enter? The text may "overlap" with the next subplot.
Perhaps you could simply create the axes with plt.subplots:
import numpy as np
import matplotlib.pyplot as plt
fig, axs = plt.subplots(ncols=2, sharey=True)
for ax in axs:
ax.plot([0,1],[0,1], 'r-')
ax.set_xticks([0.5,1.0])
ax.set_xticklabels(['0.5','1.0'])
axs[-1].legend(('Some plot'), loc=2)
for ax in axs[1:]:
ax.yaxis.set_visible(False)
fig.subplots_adjust(wspace=0)
plt.show()
The code below produces gaps between the subplots. How do I remove the gaps between the subplots and make the image a tight grid?
import matplotlib.pyplot as plt
for i in range(16):
i = i + 1
ax1 = plt.subplot(4, 4, i)
plt.axis('on')
ax1.set_xticklabels([])
ax1.set_yticklabels([])
ax1.set_aspect('equal')
plt.subplots_adjust(wspace=None, hspace=None)
plt.show()
The problem is the use of aspect='equal', which prevents the subplots from stretching to an arbitrary aspect ratio and filling up all the empty space.
Normally, this would work:
import matplotlib.pyplot as plt
ax = [plt.subplot(2,2,i+1) for i in range(4)]
for a in ax:
a.set_xticklabels([])
a.set_yticklabels([])
plt.subplots_adjust(wspace=0, hspace=0)
The result is this:
However, with aspect='equal', as in the following code:
import matplotlib.pyplot as plt
ax = [plt.subplot(2,2,i+1) for i in range(4)]
for a in ax:
a.set_xticklabels([])
a.set_yticklabels([])
a.set_aspect('equal')
plt.subplots_adjust(wspace=0, hspace=0)
This is what we get:
The difference in this second case is that you've forced the x- and y-axes to have the same number of units/pixel. Since the axes go from 0 to 1 by default (i.e., before you plot anything), using aspect='equal' forces each axis to be a square. Since the figure is not a square, pyplot adds in extra spacing between the axes horizontally.
To get around this problem, you can set your figure to have the correct aspect ratio. We're going to use the object-oriented pyplot interface here, which I consider to be superior in general:
import matplotlib.pyplot as plt
fig = plt.figure(figsize=(8,8)) # Notice the equal aspect ratio
ax = [fig.add_subplot(2,2,i+1) for i in range(4)]
for a in ax:
a.set_xticklabels([])
a.set_yticklabels([])
a.set_aspect('equal')
fig.subplots_adjust(wspace=0, hspace=0)
Here's the result:
You can use gridspec to control the spacing between axes. There's more information here.
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
plt.figure(figsize = (4,4))
gs1 = gridspec.GridSpec(4, 4)
gs1.update(wspace=0.025, hspace=0.05) # set the spacing between axes.
for i in range(16):
# i = i + 1 # grid spec indexes from 0
ax1 = plt.subplot(gs1[i])
plt.axis('on')
ax1.set_xticklabels([])
ax1.set_yticklabels([])
ax1.set_aspect('equal')
plt.show()
Without resorting gridspec entirely, the following might also be used to remove the gaps by setting wspace and hspace to zero:
import matplotlib.pyplot as plt
plt.clf()
f, axarr = plt.subplots(4, 4, gridspec_kw = {'wspace':0, 'hspace':0})
for i, ax in enumerate(f.axes):
ax.grid('on', linestyle='--')
ax.set_xticklabels([])
ax.set_yticklabels([])
plt.show()
plt.close()
Resulting in:
With recent matplotlib versions you might want to try Constrained Layout. This does (or at least did) not work with plt.subplot() however, so you need to use plt.subplots() instead:
fig, axs = plt.subplots(4, 4, constrained_layout=True)
Have you tried plt.tight_layout()?
with plt.tight_layout()
without it:
Or: something like this (use add_axes)
left=[0.1,0.3,0.5,0.7]
width=[0.2,0.2, 0.2, 0.2]
rectLS=[]
for x in left:
for y in left:
rectLS.append([x, y, 0.2, 0.2])
axLS=[]
fig=plt.figure()
axLS.append(fig.add_axes(rectLS[0]))
for i in [1,2,3]:
axLS.append(fig.add_axes(rectLS[i],sharey=axLS[-1]))
axLS.append(fig.add_axes(rectLS[4]))
for i in [1,2,3]:
axLS.append(fig.add_axes(rectLS[i+4],sharex=axLS[i],sharey=axLS[-1]))
axLS.append(fig.add_axes(rectLS[8]))
for i in [5,6,7]:
axLS.append(fig.add_axes(rectLS[i+4],sharex=axLS[i],sharey=axLS[-1]))
axLS.append(fig.add_axes(rectLS[12]))
for i in [9,10,11]:
axLS.append(fig.add_axes(rectLS[i+4],sharex=axLS[i],sharey=axLS[-1]))
If you don't need to share axes, then simply axLS=map(fig.add_axes, rectLS)
Another method is to use the pad keyword from plt.subplots_adjust(), which also accepts negative values:
import matplotlib.pyplot as plt
ax = [plt.subplot(2,2,i+1) for i in range(4)]
for a in ax:
a.set_xticklabels([])
a.set_yticklabels([])
plt.subplots_adjust(pad=-5.0)
Additionally, to remove the white at the outer fringe of all subplots (i.e. the canvas), always save with plt.savefig(fname, bbox_inches="tight").