I want to annotate the bars in a graph with some text but if the bars are close together and have comparable height, the annotations are above ea. other and thus hard to read (the coordinates for the annotations were taken from the bar position and height).
Is there a way to shift one of them if there is a collision?
Edit: The bars are very thin and very close sometimes so just aligning vertically doesn't solve the problem...
A picture might clarify things:
I've written a quick solution, which checks each annotation position against default bounding boxes for all the other annotations. If there is a collision it changes its position to the next available collision free place. It also puts in nice arrows.
For a fairly extreme example, it will produce this (none of the numbers overlap):
Instead of this:
Here is the code:
import numpy as np
import matplotlib.pyplot as plt
from numpy.random import *
def get_text_positions(x_data, y_data, txt_width, txt_height):
a = zip(y_data, x_data)
text_positions = y_data.copy()
for index, (y, x) in enumerate(a):
local_text_positions = [i for i in a if i[0] > (y - txt_height)
and (abs(i[1] - x) < txt_width * 2) and i != (y,x)]
if local_text_positions:
sorted_ltp = sorted(local_text_positions)
if abs(sorted_ltp[0][0] - y) < txt_height: #True == collision
differ = np.diff(sorted_ltp, axis=0)
a[index] = (sorted_ltp[-1][0] + txt_height, a[index][1])
text_positions[index] = sorted_ltp[-1][0] + txt_height
for k, (j, m) in enumerate(differ):
#j is the vertical distance between words
if j > txt_height * 2: #if True then room to fit a word in
a[index] = (sorted_ltp[k][0] + txt_height, a[index][1])
text_positions[index] = sorted_ltp[k][0] + txt_height
break
return text_positions
def text_plotter(x_data, y_data, text_positions, axis,txt_width,txt_height):
for x,y,t in zip(x_data, y_data, text_positions):
axis.text(x - txt_width, 1.01*t, '%d'%int(y),rotation=0, color='blue')
if y != t:
axis.arrow(x, t,0,y-t, color='red',alpha=0.3, width=txt_width*0.1,
head_width=txt_width, head_length=txt_height*0.5,
zorder=0,length_includes_head=True)
Here is the code producing these plots, showing the usage:
#random test data:
x_data = random_sample(100)
y_data = random_integers(10,50,(100))
#GOOD PLOT:
fig2 = plt.figure()
ax2 = fig2.add_subplot(111)
ax2.bar(x_data, y_data,width=0.00001)
#set the bbox for the text. Increase txt_width for wider text.
txt_height = 0.04*(plt.ylim()[1] - plt.ylim()[0])
txt_width = 0.02*(plt.xlim()[1] - plt.xlim()[0])
#Get the corrected text positions, then write the text.
text_positions = get_text_positions(x_data, y_data, txt_width, txt_height)
text_plotter(x_data, y_data, text_positions, ax2, txt_width, txt_height)
plt.ylim(0,max(text_positions)+2*txt_height)
plt.xlim(-0.1,1.1)
#BAD PLOT:
fig = plt.figure()
ax = fig.add_subplot(111)
ax.bar(x_data, y_data, width=0.0001)
#write the text:
for x,y in zip(x_data, y_data):
ax.text(x - txt_width, 1.01*y, '%d'%int(y),rotation=0)
plt.ylim(0,max(text_positions)+2*txt_height)
plt.xlim(-0.1,1.1)
plt.show()
Another option using my library adjustText, written specially for this purpose (https://github.com/Phlya/adjustText). I think it's probably significantly slower that the accepted answer (it slows down considerably with a lot of bars), but much more general and configurable.
from adjustText import adjust_text
np.random.seed(2017)
x_data = np.random.random_sample(100)
y_data = np.random.random_integers(10,50,(100))
f, ax = plt.subplots(dpi=300)
bars = ax.bar(x_data, y_data, width=0.001, facecolor='k')
texts = []
for x, y in zip(x_data, y_data):
texts.append(plt.text(x, y, y, horizontalalignment='center', color='b'))
adjust_text(texts, add_objects=bars, autoalign='y', expand_objects=(0.1, 1),
only_move={'points':'', 'text':'y', 'objects':'y'}, force_text=0.75, force_objects=0.1,
arrowprops=dict(arrowstyle="simple, head_width=0.25, tail_width=0.05", color='r', lw=0.5, alpha=0.5))
plt.show()
If we allow autoalignment along x axis, it gets even better (I just need to resolve a small issue that it doesn't like putting labels above the points and not a bit to the side...).
np.random.seed(2017)
x_data = np.random.random_sample(100)
y_data = np.random.random_integers(10,50,(100))
f, ax = plt.subplots(dpi=300)
bars = ax.bar(x_data, y_data, width=0.001, facecolor='k')
texts = []
for x, y in zip(x_data, y_data):
texts.append(plt.text(x, y, y, horizontalalignment='center', size=7, color='b'))
adjust_text(texts, add_objects=bars, autoalign='xy', expand_objects=(0.1, 1),
only_move={'points':'', 'text':'y', 'objects':'y'}, force_text=0.75, force_objects=0.1,
arrowprops=dict(arrowstyle="simple, head_width=0.25, tail_width=0.05", color='r', lw=0.5, alpha=0.5))
plt.show()
(I had to adjust some parameters here, of course)
One option is to rotate the text/annotation, which is set by the rotation keyword/property. In the following example, I rotate the text 90 degrees to guarantee that it wont collide with the neighboring text. I also set the va (short for verticalalignment) keyword, so that the text is presented above the bar (above the point that I use to define the text):
import matplotlib.pyplot as plt
data = [10, 8, 8, 5]
fig = plt.figure()
ax = fig.add_subplot(111)
ax.bar(range(4),data)
ax.set_ylim(0,12)
# extra .4 is because it's half the default width (.8):
ax.text(1.4,8,"2nd bar",rotation=90,va='bottom')
ax.text(2.4,8,"3nd bar",rotation=90,va='bottom')
plt.show()
The result is the following figure:
Determining programmatically if there are collisions between various annotations is a trickier process. This might be worth a separate question: Matplotlib text dimensions.
Just thought I would provide an alternative solution that I just created textalloc that makes sure that text-boxes avoids overlap with both each other and lines when possible, and is fast.
For this example you could use something like this:
import textalloc as ta
import numpy as np
import matplotlib.pyplot as plt
np.random.seed(2017)
x_data = np.random.random_sample(100)
y_data = np.random.random_integers(10,50,(100))
f, ax = plt.subplots(dpi=200)
bars = ax.bar(x_data, y_data, width=0.002, facecolor='k')
ta.allocate_text(f,ax,x_data,y_data,
[str(yy) for yy in list(y_data)],
x_lines=[np.array([xx,xx]) for xx in list(x_data)],
y_lines=[np.array([0,yy]) for yy in list(y_data)],
textsize=8,
margin=0.004,
min_distance=0.005,
linewidth=0.7,
textcolor="b")
plt.show()
This results in this
Related
I have an issue with customizing the legend of my plot. I did lot's of customizing but couldnt get my head around this one. I want the symbols (not the labels) to be equally spaced in the legend. As you can see in the example, the space between the circles in the legend, gets smaller as the circles get bigger.
any ideas?
Also, how can I also add a color bar (in addition to the size), with smaller circles being light red (for example) and bigger circle being blue (for example)
here is my code so far:
import pandas as pd
import matplotlib.pyplot as plt
from vega_datasets import data as vega_data
gap = pd.read_json(vega_data.gapminder.url)
df = gap.loc[gap['year'] == 2000]
fig, ax = plt.subplots(1, 1,figsize=[14,12])
ax=ax.scatter(df['life_expect'], df['fertility'],
s = df['pop']/100000,alpha=0.7, edgecolor="black",cmap="viridis")
plt.xlabel("X")
plt.ylabel("Y");
kw = dict(prop="sizes", num=6, color="lightgrey", markeredgecolor='black',markeredgewidth=2)
plt.legend(*ax.legend_elements(**kw),bbox_to_anchor=(1, 0),frameon=False,
loc="lower left",markerscale=1,ncol=1,borderpad=2,labelspacing=4,handletextpad=2)
plt.grid()
plt.show()
It's a bit tricky, but you could measure the legend elements and reposition them to have a constant inbetween distance. Due to the pixel positioning, the plot can't be resized afterwards.
I tested the code inside PyCharm with the 'Qt5Agg' backend. And in a Jupyter notebook, both with %matplotlib inline and with %matplotlib notebook. I'm not sure whether it would work well in all environments.
Note that ax.scatter doesn't return an ax (countrary to e.g. sns.scatterplot) but a list of the created scatter dots.
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.transforms import IdentityTransform
from vega_datasets import data as vega_data
gap = pd.read_json(vega_data.gapminder.url)
df = gap.loc[gap['year'] == 2000]
fig, ax = plt.subplots(1, 1, figsize=[14, 12])
fig.subplots_adjust(right=0.8)
scat = ax.scatter(df['life_expect'], df['fertility'],
s=df['pop'] / 100000, alpha=0.7, edgecolor="black", cmap="viridis")
plt.xlabel("X")
plt.ylabel("Y")
x = 1.1
y = 0.1
is_first = True
kw = dict(prop="sizes", num=6, color="lightgrey", markeredgecolor='black', markeredgewidth=2)
handles, labels = scat.legend_elements(**kw)
inverted_transData = ax.transData.inverted()
for handle, label in zip(handles[::-1], labels[::-1]):
plt.setp(handle, clip_on=False)
for _ in range(1 if is_first else 2):
plt.setp(handle, transform=ax.transAxes)
if is_first:
xd, yd = x, y
else:
xd, yd = inverted_transData.transform((x, y))
handle.set_xdata([xd])
handle.set_ydata([yd])
ax.add_artist(handle)
bbox = handle.get_window_extent(fig.canvas.get_renderer())
y += y - bbox.y0 + 15 # 15 pixels inbetween
x = (bbox.x0 + bbox.x1) / 2
if is_first:
xd_text, _ = inverted_transData.transform((bbox.x1+10, y))
ax.text(xd_text, yd, label, transform=ax.transAxes, ha='left', va='center')
y = bbox.y1
is_first = False
plt.show()
I am trying to "fold" an exponential plot (and a fit to it - see the first image below) around a discrete interval on the x-axis (a.k.a a "modulo plot"). The aim is that after 10 x-units the exponential is continued on the same plot from 0 for the 10 to 20 interval, as shown on a second "photoshopped" image below.
The MWE code is below:
import numpy as np
from scipy import optimize
import matplotlib.pyplot as plt
Generate points
x=np.arange(20)
y=np.exp(-x/10)
Fit to data
def fit_func(x, t):
return np.exp(-x/t)
par, pcov = optimize.curve_fit(f=fit_func, xdata=x, ydata=y)
Plot data and fit function
fig, ax = plt.subplots()
ax.plot(x,y, c='g', label="Data");
ax.plot(x,fit_func(x, par), c='r', linestyle=":", label="Fit");
ax.set_xlabel("x (modulo 10)")
ax.legend()
plt.savefig("fig/mod.png", dpi=300)
What I have: Origianl exponential from 0 to 20
What I want: Modulo/folded exponential in intervals of 10
You could try to simply write:
ax.plot(x % 10,y, c='g', label="Data")
ax.plot(x % 10, f, c='r', linestyle=":", label="Fit")
but then you get confusing lines connecting the last point of one section to the first point of the next.
Another idea is to create a loop to plot every part separately. To avoid multiple legend entries, only the first section sets a legend label.
import numpy as np
from scipy import optimize
import matplotlib.pyplot as plt
x=np.arange(40)
y=np.exp(-x/10)
def fit_func(x, t):
return np.exp(-x/t)
par, pcov = optimize.curve_fit(f=fit_func, xdata=x, ydata=y)
f = fit_func(x, par)
fig, ax = plt.subplots()
left = x.min()
section = 1
while left < x.max():
right = left+10
filter = (x >= left) & (x <= right)
ax.plot(x[filter]-left,y[filter], c='g', label="Data" if section == 1 else '')
ax.plot(x[filter]-left, f[filter], c='r', linestyle=":", label="Fit" if section == 1 else '')
left = right
section += 1
ax.set_xlabel("x (modulo 10)")
ax.legend()
#plt.savefig("fig/mod.png", dpi=300)
plt.show()
Assuming that x is a sorted array, we'll have :
>>> y_ = fit_func(x, par)
>>> temp_x = []
>>> temp_y = []
>>> temp_y_ = []
>>> fig, ax = plt.subplots()
>>> for i in range(len(x)):
if x[i]%10==0 or i == len(x)-1:
ax.plot(temp_x,temp_y, c='g', label="Data");
ax.plot(temp_x,temp_y_, c='r', linestyle=":", label="Fit")
temp_x,temp_y,temp_y_ = [],[],[]
else:
temp_x.append(x[i]%10)
temp_y.append(y[i])
temp_y_.append(y_[i])
>>> plt.show()
and this would be the resulting plot :
I want to create a plot for two different datasets similar to the one presented in this answer:
In the above image, the author managed to fix the overlapping problem of the error bars by adding some small random scatter in x to the new dataset.
In my problem, I must plot a similar graphic, but having some categorical data in the x axis:
Any ideas on how to slightly move one the error bars of the second dataset using categorical variables at the x axis? I want to avoid the overlapping between the bars for making the visualization easier.
You can translate each errorbar by adding the default data transform to a prior translation in data space. This is possible when knowing that categories are in general one data unit away from each other.
import numpy as np; np.random.seed(42)
import matplotlib.pyplot as plt
from matplotlib.transforms import Affine2D
x = list("ABCDEF")
y1, y2 = np.random.randn(2, len(x))
yerr1, yerr2 = np.random.rand(2, len(x))*4+0.3
fig, ax = plt.subplots()
trans1 = Affine2D().translate(-0.1, 0.0) + ax.transData
trans2 = Affine2D().translate(+0.1, 0.0) + ax.transData
er1 = ax.errorbar(x, y1, yerr=yerr1, marker="o", linestyle="none", transform=trans1)
er2 = ax.errorbar(x, y2, yerr=yerr2, marker="o", linestyle="none", transform=trans2)
plt.show()
Alternatively, you could translate the errorbars after applying the data transform and hence move them in units of points.
import numpy as np; np.random.seed(42)
import matplotlib.pyplot as plt
from matplotlib.transforms import ScaledTranslation
x = list("ABCDEF")
y1, y2 = np.random.randn(2, len(x))
yerr1, yerr2 = np.random.rand(2, len(x))*4+0.3
fig, ax = plt.subplots()
trans1 = ax.transData + ScaledTranslation(-5/72, 0, fig.dpi_scale_trans)
trans2 = ax.transData + ScaledTranslation(+5/72, 0, fig.dpi_scale_trans)
er1 = ax.errorbar(x, y1, yerr=yerr1, marker="o", linestyle="none", transform=trans1)
er2 = ax.errorbar(x, y2, yerr=yerr2, marker="o", linestyle="none", transform=trans2)
plt.show()
While results look similar in both cases, they are fundamentally different. You will observe this difference when interactively zooming the axes or changing the figure size.
Consider the following approach to highlight plots - combination of errorbar and fill_between with non-zero transparency:
import random
import matplotlib.pyplot as plt
# create sample data
N = 8
data_1 = {
'x': list(range(N)),
'y': [10. + random.random() for dummy in range(N)],
'yerr': [.25 + random.random() for dummy in range(N)]}
data_2 = {
'x': list(range(N)),
'y': [10.25 + .5 * random.random() for dummy in range(N)],
'yerr': [.5 * random.random() for dummy in range(N)]}
# plot
plt.figure()
# only errorbar
plt.subplot(211)
for data in [data_1, data_2]:
plt.errorbar(**data, fmt='o')
# errorbar + fill_between
plt.subplot(212)
for data in [data_1, data_2]:
plt.errorbar(**data, alpha=.75, fmt=':', capsize=3, capthick=1)
data = {
'x': data['x'],
'y1': [y - e for y, e in zip(data['y'], data['yerr'])],
'y2': [y + e for y, e in zip(data['y'], data['yerr'])]}
plt.fill_between(**data, alpha=.25)
Result:
Threre is example on lib site: https://matplotlib.org/stable/gallery/lines_bars_and_markers/errorbar_subsample.html
enter image description here
You need parameter errorevery=(m, n),
n - how often plot error lines, m - shift with range from 0 to n
I am using an artist animation method with 5 subplots. There is one static plot on the left, with 3 smaller animated imshow plots to the right (the colorbar is the 5th). I have successfully used ConnectionPatch to connect subplots to show where the data is coming from, but only on static plots. No matter what I try, I can't seem to get the patches to show up on the animation. I've tried to include the patch in the image artist list, tried to update the figure with the artist instead of the axis (which I guess doesn't make much sense), among other things. It will be very difficult to extract a working example due to the complexity of the plot, but maybe someone has a tip.
Could setting the facecolor to 'white' with the animation savefig_kwargs be covering up the connector lines? If so, how do I change the z order of the patch/facecolor?
Without a minimal working example, I can only tell you that it is possible to use a ConnectionPatch in an animation. However, as seen below, one has to recreate it for every frame.
import matplotlib.pyplot as plt
import numpy as np; np.random.seed(0)
import matplotlib.gridspec as gridspec
from matplotlib.patches import ConnectionPatch
import matplotlib.animation
plt.rcParams["figure.figsize"] = np.array([6,3.6])*0.7
x = np.linspace(-3,3)
X,Y = np.meshgrid(x,x)
f = lambda x,y: (1 - x / 2. + x ** 5 + y ** 3) * np.exp(-x ** 2 - y ** 2)+1.5
Z = f(X,Y)
bins=np.linspace(Z.min(), Z.max(), 16)
cols = plt.cm.PuOr((bins[:-1]-Z.min())/(Z.max()-Z.min()))
gs = gridspec.GridSpec(2, 2, height_ratios=[34,53], width_ratios=[102,53])
fig = plt.figure()
ax=fig.add_subplot(gs[:,0])
ax2=fig.add_subplot(gs[0,1])
ax3=fig.add_subplot(gs[1,1])
ax.imshow(Z, cmap="PuOr")
rec = plt.Rectangle([-.5,-.5], width=9, height=9, edgecolor="crimson", fill=False, lw=2)
conp = ConnectionPatch(xyA=[-0.5,0.5], xyB=[9.5,4], coordsA="data", coordsB="data",
axesA=ax3, axesB=ax, arrowstyle="-|>", zorder=25, shrinkA=0, shrinkB=1,
mutation_scale=20, fc="w", ec="crimson", lw=2)
ax3.add_artist(conp)
ax.add_artist(rec)
im = ax3.imshow(Z[:9,:9], cmap="PuOr", vmin=Z.min(), vmax=Z.max())
ticks = np.array([0,4,8])
ax3.set_yticks(ticks); ax3.set_xticks(ticks)
ax2.hist(Z[:9,:9].flatten(), bins=bins)
def ins(px,py):
global rec, conp, histpatches
ll = [px-.5,py-.5]
rec.set_xy(ll)
conp.remove()
conp = ConnectionPatch(xyA=[-0.5,0.5], xyB=[px+9.5,py+4], coordsA="data", coordsB="data",
axesA=ax3, axesB=ax, arrowstyle="-|>", zorder=25, shrinkA=0, shrinkB=1,
mutation_scale=20, fc="w", ec="crimson", lw=2)
ax3.add_patch(conp)
data = Z[px:px+9,py:py+9]
im.set_data(data)
ax3.set_xticklabels(ticks+px)
ax3.set_yticklabels(ticks+py)
ax2.clear()
ax2.set_ylim(0,60)
h, b_, patches = ax2.hist(data.flatten(), bins=bins, ec="k", fc="#f1a142")
[pat.set_color(cols[i]) for i, pat in enumerate(patches)]
def func(p):
px,py = p
ins(px, py)
phi = np.linspace(0.,2*np.pi)
r = np.sin(2*phi)*20+np.pi/2
xr = (r*np.cos(phi)).astype(np.int8)
yr = (r*np.sin(phi)).astype(np.int8)
plt.subplots_adjust(top=0.93,bottom=0.11,left=0.04,right=0.96,hspace=0.26,wspace=0.15)
frames = np.c_[xr+20, yr+20]
ani = matplotlib.animation.FuncAnimation(fig, func, frames=frames, interval=300, repeat=True)
plt.show()
It's possible to fill between lines with a color:
http://matplotlib.sourceforge.net/examples/pylab_examples/fill_between_demo.html
It's also possible to use a continuous colormap for a line:
http://matplotlib.sourceforge.net/examples/pylab_examples/multicolored_line.html
Is it possible (and reasonably easy) to use a continuous colormap for the colored fill between two lines? For example, the color fill may change along x based on the difference between the two lines at x (or based on another set of data).
I found a solution to this problem. It builds on the brilliant but hacky solution of #Hooked. You create a 2D grid filed from lots of small boxes. It's not the fastest solution but it should be pretty flexible (more so than solutions which apply imshow to the patches).
import numpy as np
import pylab as plt
#Plot a rectangle
def rect(ax, x, y, w, h, c,**kwargs):
#Varying only in x
if len(c.shape) is 1:
rect = plt.Rectangle((x, y), w, h, color=c, ec=c,**kwargs)
ax.add_patch(rect)
#Varying in x and y
else:
#Split into a number of bins
N = c.shape[0]
hb = h/float(N); yl = y
for i in range(N):
yl += hb
rect = plt.Rectangle((x, yl), w, hb,
color=c[i,:], ec=c[i,:],**kwargs)
ax.add_patch(rect)
#Fill a contour between two lines
def rainbow_fill_between(ax, X, Y1, Y2, colors=None,
cmap=plt.get_cmap("Reds"),**kwargs):
plt.plot(X,Y1,lw=0) # Plot so the axes scale correctly
dx = X[1]-X[0]
N = X.size
#Pad a float or int to same size as x
if (type(Y2) is float or type(Y2) is int):
Y2 = np.array([Y2]*N)
#No colors -- specify linear
if colors is None:
colors = []
for n in range(N):
colors.append(cmap(n/float(N)))
#Varying only in x
elif len(colors.shape) is 1:
colors = cmap((colors-colors.min())
/(colors.max()-colors.min()))
#Varying only in x and y
else:
cnp = np.array(colors)
colors = np.empty([colors.shape[0],colors.shape[1],4])
for i in range(colors.shape[0]):
for j in range(colors.shape[1]):
colors[i,j,:] = cmap((cnp[i,j]-cnp[:,:].min())
/(cnp[:,:].max()-cnp[:,:].min()))
colors = np.array(colors)
#Create the patch objects
for (color,x,y1,y2) in zip(colors,X,Y1,Y2):
rect(ax,x,y2,dx,y1-y2,color,**kwargs)
# Some Test data
X = np.linspace(0,10,100)
Y1 = .25*X**2 - X
Y2 = X
g = np.exp(-.3*(X-5)**2)
#Plot fill and curves changing in x only
fig, axs =plt.subplots(1,2)
colors = g
rainbow_fill_between(axs[0],X,Y1,Y2,colors=colors)
axs[0].plot(X,Y1,'k-',lw=4)
axs[0].plot(X,Y2,'k-',lw=4)
#Plot fill and curves changing in x and y
colors = np.outer(g,g)
rainbow_fill_between(axs[1],X,Y1,Y2,colors=colors)
axs[1].plot(X,Y1,'k-',lw=4)
axs[1].plot(X,Y2,'k-',lw=4)
plt.show()
The result is,
Your solution is great and flexible ! In particular the 2D case is really nice. Such a feature could be added to fill_between maybe if the colors kwargs of the function would accept an array of the same length of x and y ?
Here is a simpler case for the 1D case using the fill_between function. It does the same but as it use trapezes instead of rectangle the result is smoother.
import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
from scipy.stats import norm
# Select a color map
cmap = mpl.cm.bwr
# Some Test data
npts = 100
x = np.linspace(-4, 4, npts)
y = norm.pdf(x)
z = np.sin(2 * x)
normalize = mpl.colors.Normalize(vmin=z.min(), vmax=z.max())
# The plot
fig = plt.figure()
ax = fig.add_axes([0.12, 0.12, 0.68, 0.78])
plt.plot(x, y, color="gray")
for i in range(npts - 1):
plt.fill_between([x[i], x[i+1]], [y[i], y[i+1]], color=cmap(normalize(z[i])))
cbax = fig.add_axes([0.85, 0.12, 0.05, 0.78])
cb = mpl.colorbar.ColorbarBase(cbax, cmap=cmap, norm=normalize, orientation='vertical')
cb.set_label("Sin function", rotation=270, labelpad=15)
plt.show()