Defining colors of Matplotlib 3D bar plot - python

I can't figure out the right way to set a cmap (or colors) for a 3d bar plot in matplotlib in my iPython notebook. I can setup my chart correctly (28 x 7 labels) in the X and Y plane, with some random Z values. The graph is hard to interpret, and one reason is that the default colors for the x_data labels [1,2,3,4,5] are all the same.
Here is the code:
%matplotlib inline
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as npfig = plt.figure(figsize=(18,12))
ax = fig.add_subplot(111, projection='3d')
x_data, y_data = np.meshgrid(np.arange(5),np.arange(3))
z_data = np.random.rand(3,5).flatten()
ax.bar3d(x_data.flatten(),
y_data.flatten(),np.zeros(len(z_data)),1,1,z_data,alpha=0.10)
Which produces the following chart:
I don't want to define the colors manually for the labels x_data. How can I set up different 'random' cmap colors for each of the labels in x_data, still keeping the
ax.bar3d
parameter? Below is a variation using
ax.bar
and different colors, but what I need is ax.bar3d.

from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
fig = plt.figure(figsize=(18,12))
ax = fig.add_subplot(111, projection='3d')
x_data, y_data = np.meshgrid(np.arange(5),np.arange(3))
z_data = np.random.rand(3,5)
colors = ['r','g','b'] # colors for every line of y
# plot colored 3d bars
for i in xrange(3): # cycle though y
# I multiply one color by len of x (it is 5) to set one color for y line
ax.bar3d(x_data[i], y_data[i], z_data[i], 1, 1, z_data[i], alpha=0.1, color=colors[i]*5)
# or use random colors
# ax.bar3d(x_data[i], y_data[i], z_data[i], 1, 1, z_data[i], alpha=0.1, color=[np.random.rand(3,1),]*5)
plt.show()
Result:

Related

matplotlib plot_surface 3D plot with non-linear color map

I have this following python code, which displays the following 3D plot.
My code is:
from mpl_toolkits.mplot3d import axes3d
import matplotlib.pyplot as plt
from matplotlib import cm
import numpy as np
# Generate data example
X,Y = np.meshgrid(np.arange(-99,-90), np.arange(-200,250,50))
Z = np.zeros_like(X)
Z[:,0] = 100.
Z[4][7] = 10
# Normalize to [0,1]
Z = (Z-Z.min())/(Z.max()-Z.min())
colors = cm.viridis(Z)
rcount, ccount, _ = colors.shape
fig = plt.figure()
ax = fig.gca(projection='3d')
surf = ax.plot_surface(X, Y, Z, rcount=rcount, ccount=ccount,
facecolors=colors, shade=False)
surf.set_facecolor((0,0,0,0))
plt.show()
I want to color the irregularities on the XY plane in a different color. I want to be able to highlight the bumps on the XY plane.
How do I do that?
The problem is that the grid is not very dense. The bump consist of a single pixel. So there are 4 cells in the grid, 3 of which have their lower left corner at 0, and would hence not receive a different color according to their value. Only the one pixel which actually is the bump gets colorized.
from mpl_toolkits.mplot3d import axes3d
import matplotlib.pyplot as plt
from matplotlib import cm
import numpy as np
X,Y = np.meshgrid(np.arange(-99,-90), np.arange(-200,250,50))
Z = np.zeros_like(X)
Z[:,0] = 100.
Z[4][7] = 10
norm = plt.Normalize(Z.min(),Z.min()+10 )
colors = cm.viridis(norm(Z))
fig = plt.figure()
ax = fig.gca(projection='3d')
surf = ax.plot_surface(X, Y, Z, facecolors=colors, shade=False)
surf.set_facecolor((0,0,0,0))
plt.show()
Now you may expand the colorized part of the plot, e.g. using scipy.ndimage.grey_dilation, such that all pixels that are adjacent also become yellow.
from scipy import ndimage
C = ndimage.grey_dilation(Z, size=(2,2), structure=np.ones((2, 2)))
norm = plt.Normalize(Z.min(),Z.min()+10 )
colors = cm.viridis(norm(C))

Superimpose independent plots in python

I want to work with only one figure, with multiples, different and modifiable plots, whithout the subplots formalism.
Is there a way to superimpose two differents plots, in the same way as text boxes, i.e anywhere on the figure ?
Here a "gimp made" example :
Thanks !
You can use figure.add_axes to place an axes at an arbitrary location.
fig = plt.figure()
fig.add_axes([0.1,0.2,0.3,0.4])
places an axes at x=0.1, y=0.2, width=0.3, height=0.4 in figure coordinates.
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
fig = plt.figure()
ax = fig.add_axes([0.4,0.1,0.5,0.6], projection='3d')
X, Y = np.meshgrid(np.arange(-5, 5, 0.25), np.arange(-5, 5, 0.25))
Z = np.sin(np.sqrt(X**2 + Y**2))
surf = ax.plot_surface(X, Y, Z, cmap="plasma")
ax = fig.add_axes([0.3,0.4,0.3,.4])
plt.plot([1,2,3])
plt.show()

Matplotlib - Plot 3D with for loop

I want to plot several 3D points with matplotlib. My coordinates are stored in 2D arrays because i got multiple cases and so i would like to plot all the cases in a same 3D plot with a "for loop" but when i do that, the results appeared on different plots...
As example :
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
X = np.array([[3,2,1],[4,5,6]])
Y = np.array([[1,2,1],[2,3,4]])
Z = np.array([[10,11,12],[13,12,16]])
for i in range(0,X.shape[0]):
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(X[i,:], Y[i,:], Z[i,:], c='r', marker='o')
ax.set_xlabel('Z')
ax.set_ylabel('X')
ax.set_zlabel('Y')
plt.show()
You create a new figure each iteration and plot it each iteration. Also you always create the first suplot of a 1x1 subplot-grid.
You probably want a x.shape[0] x 1 grid or 1 x x.shape[0] grid:
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt
import numpy as np
X = np.array([[3,2,1],[4,5,6]])
Y = np.array([[1,2,1],[2,3,4]])
Z = np.array([[10,11,12],[13,12,16]])
# Create figure outside the loop
fig = plt.figure()
for i in range(0,X.shape[0]):
# Add the i+1 subplot of the x.shape[0] x 1 grid
ax = fig.add_subplot(X.shape[0], 1, i+1, projection='3d')
ax.scatter(X[i,:], Y[i,:], Z[i,:], c='r', marker='o')
ax.set_xlabel('Z')
ax.set_ylabel('X')
ax.set_zlabel('Y')
# Show it outside the loop
plt.show()
EDIT:
If you want to plot them all into the same plot use:
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1, projection='3d')
ax.set_xlabel('Z')
ax.set_ylabel('X')
ax.set_zlabel('Y')
for i in range(0,X.shape[0]):
# Only do the scatter inside the loop
ax.scatter(X[i,:], Y[i,:], Z[i,:], c='r', marker='o')
plt.show()

Scatter plot and Color mapping in Python

I have a range of points x and y stored in numpy arrays.
Those represent x(t) and y(t) where t=0...T-1
I am plotting a scatter plot using
import matplotlib.pyplot as plt
plt.scatter(x,y)
plt.show()
I would like to have a colormap representing the time (therefore coloring the points depending on the index in the numpy arrays)
What is the easiest way to do so?
Here is an example
import numpy as np
import matplotlib.pyplot as plt
x = np.random.rand(100)
y = np.random.rand(100)
t = np.arange(100)
plt.scatter(x, y, c=t)
plt.show()
Here you are setting the color based on the index, t, which is just an array of [1, 2, ..., 100].
Perhaps an easier-to-understand example is the slightly simpler
import numpy as np
import matplotlib.pyplot as plt
x = np.arange(100)
y = x
t = x
plt.scatter(x, y, c=t)
plt.show()
Note that the array you pass as c doesn't need to have any particular order or type, i.e. it doesn't need to be sorted or integers as in these examples. The plotting routine will scale the colormap such that the minimum/maximum values in c correspond to the bottom/top of the colormap.
Colormaps
You can change the colormap by adding
import matplotlib.cm as cm
plt.scatter(x, y, c=t, cmap=cm.cmap_name)
Importing matplotlib.cm is optional as you can call colormaps as cmap="cmap_name" just as well. There is a reference page of colormaps showing what each looks like. Also know that you can reverse a colormap by simply calling it as cmap_name_r. So either
plt.scatter(x, y, c=t, cmap=cm.cmap_name_r)
# or
plt.scatter(x, y, c=t, cmap="cmap_name_r")
will work. Examples are "jet_r" or cm.plasma_r. Here's an example with the new 1.5 colormap viridis:
import numpy as np
import matplotlib.pyplot as plt
x = np.arange(100)
y = x
t = x
fig, (ax1, ax2) = plt.subplots(1, 2)
ax1.scatter(x, y, c=t, cmap='viridis')
ax2.scatter(x, y, c=t, cmap='viridis_r')
plt.show()
Colorbars
You can add a colorbar by using
plt.scatter(x, y, c=t, cmap='viridis')
plt.colorbar()
plt.show()
Note that if you are using figures and subplots explicitly (e.g. fig, ax = plt.subplots() or ax = fig.add_subplot(111)), adding a colorbar can be a bit more involved. Good examples can be found here for a single subplot colorbar and here for 2 subplots 1 colorbar.
To add to wflynny's answer above, you can find the available colormaps here
Example:
import matplotlib.cm as cm
plt.scatter(x, y, c=t, cmap=cm.jet)
or alternatively,
plt.scatter(x, y, c=t, cmap='jet')
Subplot Colorbar
For subplots with scatter, you can trick a colorbar onto your axes by building the "mappable" with the help of a secondary figure and then adding it to your original plot.
As a continuation of the above example:
import numpy as np
import matplotlib.pyplot as plt
x = np.arange(10)
y = x
t = x
fig, (ax1, ax2) = plt.subplots(1, 2)
ax1.scatter(x, y, c=t, cmap='viridis')
ax2.scatter(x, y, c=t, cmap='viridis_r')
# Build your secondary mirror axes:
fig2, (ax3, ax4) = plt.subplots(1, 2)
# Build maps that parallel the color-coded data
# NOTE 1: imshow requires a 2-D array as input
# NOTE 2: You must use the same cmap tag as above for it match
map1 = ax3.imshow(np.stack([t, t]),cmap='viridis')
map2 = ax4.imshow(np.stack([t, t]),cmap='viridis_r')
# Add your maps onto your original figure/axes
fig.colorbar(map1, ax=ax1)
fig.colorbar(map2, ax=ax2)
plt.show()
Note that you will also output a secondary figure that you can ignore.
Single colorbar for multiple subplots
sometimes it is preferable to have a single colorbar to indicate data values visualised on multiple subplots.
In this case, a Normalize() object needs to be created using the minimum and maximum data values across both plots.
Then a colorbar object can be created from a ScalarMappable() object, which maps between scalar values and colors.
import numpy as np
import matplotlib.pyplot as plt
x = np.arange(10)
y = x
t1 = x # Colour data for first plot
t2 = 2*x # Color data for second plot
all_data = np.concatenate([t1, t2])
# Create custom Normalise object using the man and max data values across both subplots to ensure colors are consistent on both plots
norm = plt.Normalize(np.min(all_data), np.max(all_data))
fig, axs = plt.subplots(1, 2)
axs[0].scatter(x, y, c=t1, cmap='viridis', norm=norm)
axs[1].scatter(x**2, y, c=t2, cmap='viridis', norm=norm)
# Create the colorbar
smap = plt.cm.ScalarMappable(cmap='viridis', norm=norm)
cbar = fig.colorbar(smap, ax=axs, fraction=0.1, shrink = 0.8)
cbar.ax.tick_params(labelsize=11)
cbar.ax.set_ylabel('T', rotation=0, labelpad = 15, fontdict = {"size":14})
plt.show()
subplots_colorbar

Python : How to plot 3d graphs using Python?

I am using matplotlib for doing this
from mpl_toolkits.mplot3d import Axes3D
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
fig = plt.figure()
ax = Axes3D(fig)
x = [6,3,6,9,12,24]
y = [3,5,78,12,23,56]
ax.plot(x, y, zs=0, zdir='z', label='zs=0, zdir=z')
plt.show()
Now this builds a graph that is horizontal in the 3d space. How do I make the graph vertical so that it faces the user?
What I want to do is build multiple such vertical graphs that are separated by some distance and are facing the user.
bp's answer might work fine, but there's a much simpler way.
Your current graph is 'flat' on the z-axis, which is why it's horizontal. You want it to be vertical, which means that you want it to be 'flat' on the y-axis. This involves the tiniest modification to your code:
from mpl_toolkits.mplot3d import Axes3D
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
fig = plt.figure()
ax = Axes3D(fig)
x = [6,3,6,9,12,24]
y = [3,5,78,12,23,56]
# put 0s on the y-axis, and put the y axis on the z-axis
ax.plot(xs=x, ys=[0]*len(x), zs=y, zdir='z', label='ys=0, zdir=z')
plt.show()
Then you can easily have multiple such graphs by using different values for the ys parameter (for example, ys=[2]*len(x) instead would put the graph slightly behind).
Mayavi, in particular the mlab module, provides powerful 3D plotting that will work on large and or complex data, and should be easy to use on numpy arrays.
You can set the view angle of the 3d plot with the view_init() function. The example below is for version 1.1 of matplotlib.
from mpl_toolkits.mplot3d import axes3d
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
x = [6,3,6,9,12,24]
y = [3,5,78,12,23,56]
ax.plot(x, y, zs=0, zdir='z', label='zs=0, zdir=z')
ax.view_init(90, -90)
plt.show()
According to the documentation you want to use the ax.plot_surface(x,y,z) method. More information and chart types here.
The following should work:
x = [1,2,3]
y = [4,5,6]
z = [7,8,9]
data = zip(x,y,z)
#map data on the plane
X, Y = numpy.meshgrid(arange(0, max(x), 1), arange(0, max(y), 1))
Z = numpy.zeros((len(Y), len(X)), 'Float32')
for x_,y_,z_ in data:
Z[x_, y_] = z_ #this should work, but only because x and y are integers
#and arange was done with a step of 1, starting from 0
fig = p.figure()
ax = p3.Axes3D(fig)
ax.plot_surface(X, Y, Z)

Categories

Resources