I'm trying to understand how to build arrays for use in plot_surface (in Axes3d).
I tried to build a simple surface manipulating data of those arrays:
In [106]: x
Out[106]:
array([[0, 0],
[0, 1],
[0, 0]])
In [107]: y
Out[107]:
array([[0, 0],
[1, 1],
[0, 0]])
In [108]: z
Out[108]:
array([[0, 0],
[1, 1],
[2, 2]])
But I can't figure out how they are interpreted - for example there is nothing in z=2 on my plot.
Anybody please explain exactly which values will be taken to make point, which for line and finally surface.
For example I would like to build a surface that would connect with lines points:
[0,0,0]->[1,1,1]->[0,0,2]
[0,0,0]->[1,-1,1]->[0,0,2]
and a surface between those lines.
What should arrays for plot_surface look like to get something like this?
Understanding how the grids in plot_surface work is not easy. So first I'll give a general explanation, and then I'll explain how to convert the data in your case.
If you have an array of N x values and an array of M y values, you need to create two grids of x and y values of dimension (M,N) each. Fortunately numpy.meshgrid will help. Confused? See an example:
x = np.arange(3)
y=np.arange(1,5)
X, Y = np.meshgrid(x,y)
The element (x[i], y[j]) is accessed as (X[j,i], Y[j,i]). And its Z value is, of course, Z[j,i], which you also need to define.
Having said that, your data does produce a point of the surface in (0,0,2), as expected. In fact, there are two points at that position, coming from coordinate indices (0,0,0) and (1,1,1).
I attach the result of plotting your arrays with:
fig = plt.figure()
ax=fig.add_subplot(1,1,1, projection='3d')
surf=ax.plot_surface(X, Y, Z)
If I understand you correctly you try to interpolate a surface through a set of points. I don't think the plot_surface is the correct function for this. But correct me if I'm wrong. I think you should look for interpolation tools, probably those in scipy.interpolate. The result of the interpolation can then be plotted using plot_surface.
plot_surface is able to plot a grid (with z values) in 3D space based on x, y coordinates. The arrays of x and y are those created by numpy.meshgrid.
example of plot_surface:
import pylab as plt
import numpy as np
from mpl_toolkits.mplot3d import Axes3D
plt.ion()
x = np.arange(0,np.pi, 0.1)
y = x.copy()
z = np.sin(x).repeat(32).reshape(32,32)
X, Y = np.meshgrid(x,y)
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.plot_surface(X,Y,z, cmap=plt.cm.jet, cstride=1, rstride=1)
Related
I want to make a scatterplot (using matplotlib) where the points are shaded according to a third variable. I've got very close with this:
plt.scatter(w, M, c=p, marker='s')
where w and M are the data points and p is the variable I want to shade with respect to.
However I want to do it in greyscale rather than colour. Can anyone help?
There's no need to manually set the colors. Instead, specify a grayscale colormap...
import numpy as np
import matplotlib.pyplot as plt
# Generate data...
x = np.random.random(10)
y = np.random.random(10)
# Plot...
plt.scatter(x, y, c=y, s=500) # s is a size of marker
plt.gray()
plt.show()
Or, if you'd prefer a wider range of colormaps, you can also specify the cmap kwarg to scatter. To use the reversed version of any of these, just specify the "_r" version of any of them. E.g. gray_r instead of gray. There are several different grayscale colormaps pre-made (e.g. gray, gist_yarg, binary, etc).
import matplotlib.pyplot as plt
import numpy as np
# Generate data...
x = np.random.random(10)
y = np.random.random(10)
plt.scatter(x, y, c=y, s=500, cmap='gray')
plt.show()
In matplotlib grey colors can be given as a string of a numerical value between 0-1.
For example c = '0.1'
Then you can convert your third variable in a value inside this range and to use it to color your points.
In the following example I used the y position of the point as the value that determines the color:
from matplotlib import pyplot as plt
x = [1, 2, 3, 4, 5, 6, 7, 8, 9]
y = [125, 32, 54, 253, 67, 87, 233, 56, 67]
color = [str(item/255.) for item in y]
plt.scatter(x, y, s=500, c=color)
plt.show()
Sometimes you may need to plot color precisely based on the x-value case. For example, you may have a dataframe with 3 types of variables and some data points. And you want to do following,
Plot points corresponding to Physical variable 'A' in RED.
Plot points corresponding to Physical variable 'B' in BLUE.
Plot points corresponding to Physical variable 'C' in GREEN.
In this case, you may have to write to short function to map the x-values to corresponding color names as a list and then pass on that list to the plt.scatter command.
x=['A','B','B','C','A','B']
y=[15,30,25,18,22,13]
# Function to map the colors as a list from the input list of x variables
def pltcolor(lst):
cols=[]
for l in lst:
if l=='A':
cols.append('red')
elif l=='B':
cols.append('blue')
else:
cols.append('green')
return cols
# Create the colors list using the function above
cols=pltcolor(x)
plt.scatter(x=x,y=y,s=500,c=cols) #Pass on the list created by the function here
plt.grid(True)
plt.show()
A pretty straightforward solution is also this one:
fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(8,8))
p = ax.scatter(x, y, c=y, cmap='cmo.deep')
fig.colorbar(p,ax=ax,orientation='vertical',label='labelname')
In a previous question (fastest way to use numpy.interp on a 2-D array) someone asked for the fastest way to implement the following:
np.array([np.interp(X[i], x, Y[i]) for i in range(len(X))])
assume X and Y are matrices with many rows so the for loop is costly. There is a nice solution in this case that avoids the for loop (see linked answer above).
I am faced with a very similar problem, but I am unclear on whether the for loop can be avoided in this case:
np.array([np.interp(x, X[i], Y[i]) for i in range(len(X))])
In other words, I want to use linear interpolation to upsample a large number of signals stored in the rows of two matrices X and Y.
I was hoping to find a function in numpy or scipy (scipy.interpolate.interp1d) that supported this operation via broadcasting semantics but I so far can't seem to find one.
Other points:
If it helps, the rows X[i] and x are pre-sorted in my application. Also, in my case len(x) is quite a bit larger than len(X[i]).
The function scipy.signal.resample almost does what I want, but it doesn't use linear interpolation...
This is a vectorized approach that directly implements linear interpolation. First, for each x value and each i, j compute the weight w expressing how much of the interval (X[i, j], X[i, j+1]) is to the left of x.
If the entire interval is to the left of x, the weight of that interval is 1.
If none of the subinterval is to the left, the weight is 0
Otherwise, the weight is a number between 0 and 1, expressing the proportion of that interval to the left of x.
Then the value of PL interpolant is computed as Y[i, 0] + sum of differences dY[i, j] multiplied by the corresponding weight. The logic is to follow by how much the interpolant changes from interval to interval. The differences dY = np.diff(Y, axis=1) show how much it changes over the entire interval. Multiplication by the weight prorates that change accordingly.
Setup, with some small data arrays
import numpy as np
X = np.array([[0, 2, 5, 6, 9], [1, 3, 4, 7, 8]])
Y = np.array([[3, 5, 2, 4, 1], [8, 6, 9, 5, 4]])
x = np.linspace(1, 8, 20)
The computation
dX = np.diff(X, axis=1)
dY = np.diff(Y, axis=1)
w = np.clip((x - X[:, :-1, None])/dX[:, :, None], 0, 1)
y = Y[:, [0]] + np.sum(w*dY[:, :, None], axis=1)
Demonstration
This is only to show that the interpolation is correct. Blue points: original data, red ones are computed.
import matplotlib.pyplot as plt
plt.plot(x, y[0], 'ro')
plt.plot(X[0], Y[0], 'bo')
plt.plot(x, y[1], 'rd')
plt.plot(X[1], Y[1], 'bd')
plt.show()
I need to draw many such rows (for a0 .. a128) in a single window. I've searched in FacetGrid, PairGrid and all over around but couldn't find. Only regplot has similar argument ax but it doesn't plot histograms. My data is 128 real valued features with label column [0, 1]. I need the graphs to be shown from my Python code as a separate application on Linux.
Also, it there a way to scale this histogram to show relative values on Y such that the right curve is not skewed?
g = sns.FacetGrid(df, col="Result")
g.map(plt.hist, "a0", bins=20)
plt.show()
Just a simple example using matplotlib. The code is not optimized (ugly, but simple plot-indexing):
import numpy as np
import matplotlib.pyplot as plt
N = 5
data = np.random.normal(size=(N*N, 1000))
f, axarr = plt.subplots(N, N) # maybe you want sharex=True, sharey=True
pi = [0,0]
for i in range(data.shape[0]):
if pi[1] == N:
pi[0] += 1 # next row
pi[1] = 0 # first column again
axarr[pi[0], pi[1]].hist(data[i], normed=True) # i was wrong with density;
# normed=True should be used
pi[1] += 1
plt.show()
Output:
I'm plotting the degree of freedom against the square error,:
plt.plot([1,2,3,4], [0.5,0.6,0.9,0.85],'-')
It will produce
The problem is that ,the x ax is has 0.5 interval, and does not make sense in this context. Because there is simply no 1.5 degree of freedom.
How can I make the x axis into [1,2,3,4,], instead of [1, 1.5, 2, ...]?
Just add directly the positions and the strings you want to put in the x axis. Using your example:
import matplotlib.pyplot as plt
x = [1,2,3,4]
y = [0.5,0.6,0.9,0.85]
plt.plot(x,y,'-')
plt.xticks(list(range(1,max(x)+1)),[str(i) for i in range(1,max(x)+1)])
plt.grid()
plt.show()
, which results in:
You have to set the XTick 1 to 4, by 1 1:1:4 like below
plot([1,2,3,4], [0.5,0.6,0.9,0.85],'-');
set(gca,'XTick',1:1:4);
or
p = plot([1,2,3,4], [0.5,0.6,0.9,0.85],'-');
set(p,'XTick',1:1:4);
I would like to annotate the data points with their values next to the points on the plot. The examples I found only deal with x and y as vectors. However, I would like to do this for a pandas DataFrame that contains multiple columns.
ax = plt.figure().add_subplot(1, 1, 1)
df.plot(ax = ax)
plt.show()
What is the best way to annotate all the points for a multi-column DataFrame?
Here's a (very) slightly slicker version of Dan Allan's answer:
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import string
df = pd.DataFrame({'x':np.random.rand(10), 'y':np.random.rand(10)},
index=list(string.ascii_lowercase[:10]))
Which gives:
x y
a 0.541974 0.042185
b 0.036188 0.775425
c 0.950099 0.888305
d 0.739367 0.638368
e 0.739910 0.596037
f 0.974529 0.111819
g 0.640637 0.161805
h 0.554600 0.172221
i 0.718941 0.192932
j 0.447242 0.172469
And then:
fig, ax = plt.subplots()
df.plot('x', 'y', kind='scatter', ax=ax)
for k, v in df.iterrows():
ax.annotate(k, v)
Finally, if you're in interactive mode you might need to refresh the plot:
fig.canvas.draw()
Which produces:
Or, since that looks incredibly ugly, you can beautify things a bit pretty easily:
from matplotlib import cm
cmap = cm.get_cmap('Spectral')
df.plot('x', 'y', kind='scatter', ax=ax, s=120, linewidth=0,
c=range(len(df)), colormap=cmap)
for k, v in df.iterrows():
ax.annotate(k, v,
xytext=(10,-5), textcoords='offset points',
family='sans-serif', fontsize=18, color='darkslategrey')
Which looks a lot nicer:
Do you want to use one of the other columns as the text of the annotation? This is something I did recently.
Starting with some example data
In [1]: df
Out[1]:
x y val
0 -1.015235 0.840049 a
1 -0.427016 0.880745 b
2 0.744470 -0.401485 c
3 1.334952 -0.708141 d
4 0.127634 -1.335107 e
Plot the points. I plot y against x, in this example.
ax = df.set_index('x')['y'].plot(style='o')
Write a function that loops over x, y, and the value to annotate beside the point.
def label_point(x, y, val, ax):
a = pd.concat({'x': x, 'y': y, 'val': val}, axis=1)
for i, point in a.iterrows():
ax.text(point['x'], point['y'], str(point['val']))
label_point(df.x, df.y, df.val, ax)
draw()
Let's assume your df has multiple columns, and three of which are x, y, and lbl. To annotate your (x,y) scatter plot with lbl, simply:
ax = df.plot(kind='scatter',x='x',y='y')
df[['x','y','lbl']].apply(lambda row: ax.text(*row),axis=1);
I found the previous answers quite helpful, especially LondonRob's example that improved the layout a bit.
The only thing that bothered me is that I don't like pulling data out of DataFrames to then loop over them. Seems a waste of the DataFrame.
Here was an alternative that avoids the loop using .apply(), and includes the nicer-looking annotations (I thought the color scale was a bit overkill and couldn't get the colorbar to go away):
ax = df.plot('x', 'y', kind='scatter', s=50 )
def annotate_df(row):
ax.annotate(row.name, row.values,
xytext=(10,-5),
textcoords='offset points',
size=18,
color='darkslategrey')
_ = df.apply(annotate_df, axis=1)
Edit Notes
I edited my code example recently. Originally it used the same:
fig, ax = plt.subplots()
as the other posts to expose the axes, however this is unnecessary and makes the:
import matplotlib.pyplot as plt
line also unnecessary.
Also note:
If you are trying to reproduce this example and your plots don't have the points in the same place as any of ours, it may be because the DataFrame was using random values. It probably would have been less confusing if we'd used a fixed data table or a random seed.
Depending on the points, you may have to play with the xytext values to get better placements.