I read through previous solutions but couldn´t make any of them work. I want to have a global legend for individual subplots. The ax objects for this subplot are generated by a predefined function "get_plot" out of a predefined class "The_predefined_plotting_class" roughly like this:
the function returns an ax object and each ax objects has multiple "plots"/ from multiple columns of the original "datafiles".
In one of the solutions i found on this site i read that i could use:
to make a global legend. Unfortunately i have no idea how to append the individual ax objects ( or the data therein) to handles to make this work. Each plots contains some identical column names and some that differ. If an entry/name exists in many subplots it should only be printed once.
EDIT
I am really sorry that i had to use pictures, but whatever i did the webside did not let me post my code even tho it was shown properly in the preview window ( the screenshots are from this window)
EDIT2
If i do it like this:
lines=[]
labels=[]
for idata, datafile in enumerate(datafiles):
MYData = The_predefined_plotting_class.from_file(datafile)
axis[idata] = The_predefined_plotting_class.get_plot( *kwargs)
h, l = axis[idata].get_legend_handles_labels()
lines.append(h)
labels.append(l)
LINES=[]
LABELS=[]
for i in range(0, nrows):
LINES+=lines[i]
LABELS+=labels[i]
plt.legend( LINES, LABELS, loc="upper left", bbox_to_anchor=[0, 1],ncol=3, shadow=True, title="Legend", fancybox=True)
plt.show()
Then it shows all Data . Some of the Data has the same lines and labels handler. I am now left with the problem to iterate through both lists and delete only one entry if in both lists the tuple (LINES[j];LABELS[j]) = (LINES[i];LABELS[i]) exist twice (and only then). Preferably the first entry:
EDIT3
labels =[]
lines = []
h=["Cat","Mouse","Dog","Cat","Cat","Kangaroo","Dog"]
l=["black","white","brown","white","black","yellow","brown"]
for handle, label in zip(h, l):
if label not in labels :
lines.append(handle)
labels.append(label)
print "The disired Output is :"
print '["Cat","Mouse","Dog","Cat","Kangaroo"]'
print '["black","white","brown","white","yellow"]'
print "currently you get:"
print lines
print labels
EDIT4
I add a "minimum" working example which should contain all possible situations that occur in my real data.
lines=[]
labels=[]
legend_properties = {'weight':'bold','size':10}
# Example data
x1 = np.linspace(0.0, 5.0)
x2 = np.linspace(0.0, 2.0)
a = np.cos(2 * np.pi * x1) * np.exp(-x1)
b = np.cos(2 * np.pi * x2)
c = np.cos(5 * np.pi * x1) * np.exp(-x1)
c2 = np.cos(5 * np.pi * x1**2) * np.exp(-x1)
d = np.cos(2 * np.pi * x2 )
d2 = np.cos(2 * np.pi * x2-1 )
e = x1*5
e2 = -x1*5
f = np.exp(x1)-e
f2 = (np.exp(x1)-e)/2
nrows = 4
# Plot
fig, axis = plt.subplots(nrows, sharex=True, sharey=False, figsize=(5, 8))
fig.subplots_adjust(hspace=0.0001)
fig.suptitle("Stacked Plots with global Legend wich contains to little elements",fontsize=14,weight='bold')
axis[0].plot(x1, e, 'k--', label='Label1',color="green")
axis[0].plot(x1, e2, 'k--', label='Label2',color="blue")
axis[0].plot(x1, a, 'k--', label='Label3',color="yellow")
axis[1].plot(x1, c, 'k--', label='Label1',color="green")
axis[1].plot(x1, c2, 'k--', label='Label2',color="blue")
axis[1].plot(x1, a, 'k--', label='Label3',color="grey")
axis[2].plot(x2, d, '*', label='Label1',color="green")
axis[2].plot(x2, d2, 'D', label='Label2',color="green")
axis[3].plot(x1, f, 'H', label='Label1',color="green")
axis[3].plot(x1, f2, 'D', label='Label2',color="green")
for i in range(nrows):
h, l = axis[i].get_legend_handles_labels()
for handle, label in zip(h, l):
if label not in labels:
lines.append(handle)
labels.append(label)
# only 3 Legend entrys Label1 , Label2 and Label3 are visible .. Differences in cloors and markers are ignored
plt.legend(handles=lines, labels=labels,bbox_to_anchor=(0., nrows+.02, 1., .102), loc=3,ncol=3, prop=legend_properties,mode="expand", borderaxespad=0.,frameon=False,framealpha=0.0)
plt.show()
EDIT5
this is the part form the script in question where the actual plots are generated. "columns" contains just the names of the actual data to be ploted.
# add plots
ic = 0
for col in columns:
if col == "envelope":
ax.plot(self.data.index, self.data.envelope,
linewidth=LINEWIDTH_envelope, c=last_color, label="")
elif col == "Exp":
ax.plot(self.data.index, self.data.Exp, c=first_color, linestyle="",
label="Exp", marker="o", markersize=MARKERSIZE )
else:
color = used_colors[ic % len(used_colors)]
if fill and "BG" in self.data.columns:
ax.fill_between(self.data.index, self.data.BG,
self.data[col], label=col, alpha=ALPHA,
color=color)
else:
ax.plot(self.data.index, self.data[col], linewidth=LINEWIDTH,
c=color, label=col)
ic += 1
EDIT6
I tried to find a solution based on the idea i presented here:
Unfortunately what works for two Lists containing strings does not work for the artist handels it seems.
import matplotlib.pyplot as plt
import numpy as np
LI=[]
lines=[]
labels=[]
legend_properties = {'weight':'bold','size':10}
# Example data
x1 = np.linspace(0.0, 5.0)
x2 = np.linspace(0.0, 2.0)
a = np.cos(2 * np.pi * x1) * np.exp(-x1)
b = np.cos(2 * np.pi * x2)
c = np.cos(5 * np.pi * x1) * np.exp(-x1)
c2 = np.cos(5 * np.pi * x1**2) * np.exp(-x1)
d = np.cos(2 * np.pi * x2 )
d2 = np.cos(2 * np.pi * x2-1 )
e = x1*5
e2 = -x1*5
f = np.exp(x1)-e
f2 = (np.exp(x1)-e)/2
nrows = 4
# Plot
fig, axis = plt.subplots(nrows, sharex=True, sharey=False, figsize=(5, 8))
fig.subplots_adjust(hspace=0.0001)
#fig.suptitle("Stacked Plots with global Legend wich contains to little elements",fontsize=14,weight='bold')
axis[0].plot(x1, e, 'k--', label='Label1',color="green")
axis[0].plot(x1, e2, 'k--', label='Label2',color="blue")
axis[0].plot(x1, a, 'k--', label='Label3',color="yellow")
axis[1].plot(x1, c, 'k--', label='Label1',color="green")
axis[1].plot(x1, c2, 'k--', label='Label2',color="blue")
axis[1].plot(x1, a, 'k--', label='Label3',color="grey")
axis[2].plot(x2, d, '*', label='Label1',color="green")
axis[2].plot(x2, d2, 'D', label='Label2',color="green")
axis[3].plot(x1, f, 'H', label='Label1',color="green")
axis[3].plot(x1, f2, 'D', label='Label2',color="green")
for i in range(nrows):
print i
h, l = axis[i].get_legend_handles_labels()
for hl in zip(h,l):
if hl not in LI:
LI.append(hl)
lines.append(LI[-1][0])
labels.append(LI[-1][1])
print LI
# only 3 Legend entrys Label1 , Label2 and Label3 are visible .. Differences in cloors and markers are ignored
plt.legend(handles=lines, labels=labels,bbox_to_anchor=(0., nrows+.02, 1., .102), loc=3,ncol=3, prop=legend_properties,mode="expand", borderaxespad=0.,frameon=False,framealpha=0.0)
plt.show()
I think the problem is that only the string for the Memory adress is compared in
if hl not in LI:
not the actual content of "h"?
solution based on the explanation of ImportanceOfBeingErnest gave in a related post Link7:
import matplotlib.pyplot as plt
import numpy as np
import matplotlib.mlab as mlab
import math
import matplotlib.collections
def is_inlist(handle, handles):
for h in handles:
if isinstance(handle, matplotlib.collections.PolyCollection) and isinstance(h, matplotlib.collections.PolyCollection):
if np.all(h.get_facecolor() == handle.get_facecolor()) and \
np.all(h.get_linestyle() == handle.get_linestyle()) and \
np.all(h.get_alpha() == handle.get_alpha()):
return True
if isinstance(handle, matplotlib.lines.Line2D) and isinstance(h, matplotlib.lines.Line2D):
if h.get_color() == handle.get_color() and \
h.get_linestyle() == handle.get_linestyle() and \
h.get_marker() == handle.get_marker():
return True
return False
lines=[]
labels=[]
legend_properties = {'weight':'bold','size':10}
# Example data
mu = 0
mu2 = 5
variance = 1
variance2 = 2
sigma = math.sqrt(variance)
sigma2 = math.sqrt(variance2)
x = np.linspace(mu-3*variance,mu+3*variance, 100)
x2 = np.linspace(mu2-3*variance2,mu2+3*variance2, 100)
nrows = 4
# Plot
fig, axis = plt.subplots(nrows, sharex=True, sharey=False, figsize=(5, 8))
fig.subplots_adjust(hspace=0.0001)
#fig.suptitle("Stacked Plots with global Legend wich contains to little elements",fontsize=14,weight='bold')
axis[0].fill_between(x+6,0,mlab.normpdf(x, mu, sigma), color='green',alpha=0.5,label="PEAK1", interpolate=True)
axis[0].fill_between(x+4,0,mlab.normpdf(x, mu, sigma), color='orange',alpha=0.5,label="PEAK2", interpolate=True)
axis[0].fill_between(x+3,0,mlab.normpdf(x, mu, sigma), color='blue',alpha=0.5,label="PEAK3", interpolate=True)
axis[0].fill_between(x+7,0,mlab.normpdf(x, mu, sigma), color='red',alpha=0.5,label="PEAK4", interpolate=True)
axis[0].plot(x2,2.5*mlab.normpdf(x2, mu2, sigma2),color='black',linestyle="",label="Exp", marker="o", markersize=4)
axis[1].fill_between(x+6,0,mlab.normpdf(x, mu, sigma), color='green',alpha=0.5,label="PEAK1", interpolate=True)
axis[1].fill_between(x+4,0,mlab.normpdf(x, mu, sigma), color='purple',alpha=0.5,label="PEAK2", interpolate=True)
axis[1].fill_between(x+3,0,mlab.normpdf(x, mu, sigma), color='blue',alpha=0.5,label="PEAK3", interpolate=True)
axis[1].fill_between(x+7,0,mlab.normpdf(x, mu, sigma), color='red',alpha=0.5,label="PEAK4", interpolate=True)
axis[1].fill_between(x+6.5,0,mlab.normpdf(x, mu, sigma), color='yellow',alpha=0.5,label="PEAK5", interpolate=True)
axis[1].plot(x2,2.5*mlab.normpdf(x2, mu2, sigma2),color='black',linestyle="",label="Exp", marker="o", markersize=4)
axis[2].fill_between(x+6,0,mlab.normpdf(x, mu, sigma), color='green',alpha=0.5,label="PEAK1", interpolate=True)
axis[2].fill_between(x+4,0,mlab.normpdf(x, mu, sigma), color='orange',alpha=0.5,label="PEAK2", interpolate=True)
axis[2].fill_between(x+3,0,mlab.normpdf(x, mu, sigma), color='#73d216',alpha=0.5,label="PEAK3", interpolate=True)
axis[2].fill_between(x+7,0,mlab.normpdf(x, mu, sigma), color='red',alpha=0.5,label="PEAK4", interpolate=True)
axis[2].plot(x2,2.5*mlab.normpdf(x2, mu2, sigma2),color='black',linestyle="",label="Exp", marker="o", markersize=4)
axis[3].fill_between(x+6,0,mlab.normpdf(x, mu, sigma), color='green',alpha=0.5,label="PEAK1", interpolate=True)
axis[3].fill_between(x+4,0,mlab.normpdf(x, mu, sigma), color='purple',alpha=0.5,label="PEAK2", interpolate=True)
axis[3].fill_between(x+3,0,mlab.normpdf(x, mu, sigma), color='blue',alpha=0.5,label="PEAK3", interpolate=True)
axis[3].fill_between(x+7,0,mlab.normpdf(x, mu, sigma), color='red',alpha=0.5,label="PEAK4", interpolate=True)
axis[3].fill_between(x+6.5,0,mlab.normpdf(x, mu, sigma), color='#73d216',alpha=0.5,label="PEAK5", interpolate=True)
axis[3].fill_between(x+5.5,0,mlab.normpdf(x, mu, sigma), color='violet',alpha=0.5,label="PEAK6", interpolate=True)
axis[3].plot(x2,2.5*mlab.normpdf(x2, mu2, sigma2),color='black',linestyle="",label="Exp", marker="o", markersize=4)
for i in range(nrows):
h, l = axis[i].get_legend_handles_labels()
for hi, li in zip(h,l):
if not is_inlist(hi, lines):
lines.append(hi)
labels.append(li)
# only 3 Legend entrys Label1 , Label2 and Label3 are visible .. Differences in cloors and markers are ignored
plt.legend(handles=lines, labels=labels,bbox_to_anchor=(0., nrows-1+.02, 1., .102), loc=3,ncol=3, prop=legend_properties,mode="expand", borderaxespad=0.,frameon=False,framealpha=0.0)
plt.show()
Here my real data is reflected better as i have both matplotlib.collections.PolyCollection) and matplotlib.lines.Line2D objects wich need to be compared.
The Edit2 looks promising. You can then check if the label is already in the labels list and if not, append it. Of course I cannot test the following, but it should at least show the concept.
If you want to aviod diplicate handles, you may use the properties which make them appear equal and see if a similar artist is already present in the handles list.