Get color of a scatter point

Question

I have a scatter plot with some toy data.

I want to draw a label next to a given point with the color of the point.

Toy example:

x = 100*np.random.rand(5,1)
y = 100*np.random.rand(5,1)
c = np.random.rand(5,1)

fig, ax = plt.subplots()
sc = plt.scatter(x, y, c=c, cmap='viridis')

# I want to annotate the third point (idx=2)
idx = 2  
ax.annotate("hello", xy=(x[idx],y[idx]), color='green', 
            xytext=(5,5), textcoords="offset points")
plt.show()

I need to somehow get the color of this point and change my color='green' part for color=color_of_the_point

How can I get the color of a point in a scatter plot?

The color vector c is transformed to a colormap, and it can also have further modifications such as normalization or alpha value.

sc has a method for retrieving the coordinates of the points:

sc.get_offsets()

So it would be logical to also have a method for getting the color of the points but I could not find such method.

Answer 1

The scatter plot is a PathCollection, which subclasses ScalarMappable. A ScalarMappable has a method to_rgba. This can be used to get the color corresponding to the colorvalues.

In this case

sc.to_rgba(c[idx])

Note that the arrays used in the question are 2D arrays, which is normally undesired. So a complete example would look like

import matplotlib.pyplot as plt
import numpy as np

x = 100*np.random.rand(5)
y = 100*np.random.rand(5)
c = np.random.rand(5)

fig, ax = plt.subplots()
sc = plt.scatter(x, y, c=c, cmap='viridis')

# I want to annotate the third point (idx=2)
idx = 2  
ax.annotate("hello", xy=(x[idx],y[idx]), color=sc.to_rgba(c[idx]), 
            xytext=(5,5), textcoords="offset points")
plt.show()

Answer 2

The scatter plot is a PathCollection as mentioned by the other answer. It has a get_facecolors() method which returns the color used to render each point. However, the correct colors are only returned once the scatter plot is rendered. So, we can first trigger a plt.draw() and then use get_facecolors().

Working example:

import matplotlib.pyplot as plt
import numpy as np

x = 100*np.random.rand(5)
y = 100*np.random.rand(5)
c = np.random.rand(5)

fig, ax = plt.subplots()
sc = ax.scatter(x, y, c=c, cmap='viridis')

idx = 2  
plt.draw()

col = sc.get_facecolors()[idx].tolist()

ax.annotate("hello", xy=(x[idx],y[idx]), color=col, 
            xytext=(5,5), textcoords="offset points")
plt.show()

which generates