Why the figure showed by matplotlib has more than three colors while I have only three labels?

Question

I'm starting to learn to use matplotlib to draw figures. When I was using the famous iris dataset and trying to draw a plot figure, I encountered a question.

import numpy as np
import pandas as pd
import matplotlib.pylab as pl

raw = pd.read_csv('iris.csv')
data = raw.values
print data
x = data[:,0]
y = data[:,1]
pl.scatter(x,y,color = ['r','g','b'], s = [30,40,50], alpha=0.5)
pl.figure()
pl.show()
labels = set(data[:,4])
print labels

I got the output

 ...
 [6.7 3.3 5.7 2.5 'Iris-virginica']
 [6.7 3.0 5.2 2.3 'Iris-virginica']
 [6.3 2.5 5.0 1.9 'Iris-virginica']
 [6.5 3.0 5.2 2.0 'Iris-virginica']
 [6.2 3.4 5.4 2.3 'Iris-virginica']
 [5.9 3.0 5.1 1.8 'Iris-virginica']]
set(['Iris-virginica', 'Iris-setosa', 'Iris-versicolor'])

I only used the first two features because I didn't know whether it is possible to draw high dimensional figures.

This is the figure I got

There were more than three colors while, you can see from the output, there were exactly three labels ('Iris-virginica', 'Iris-setosa', 'Iris-versicolor') .

I wonder how does matplotlib decide what color to use?
What are the different colors for?

What should I do to show a three-color plot figure?

Answer 1

You obtained this figure with pyplot.scatter , more specifically with this line of code:

pl.scatter(x, y, color=['r','g','b'], s=[30,40,50], alpha=0.5)

In the line above, there is no indications whatsoever about labels. x and y are only two list of numbers.

To color the dots, scatter uses the argument color=['r', 'g', 'b'] . If color is the same size than x and y , then each dot has a defined color. But if color is smaller than x and y , then scatter will loop through color as many times as needed. For example:

x = [1, 2, 3, 4, 5]    
color = ['r', 'g', 'b'] becomes ['r', 'g', 'b', 'r', 'g']

And for the last mystery "why is there more than three colors on the plots" , it's because the transparency alpha is set to 0.5 (all colors are 50% transparent). Some of the data points have the same x and y coordinates, so the colors overlay, and it looks like there's more colors than red, green, blue.

---

To plot the right colors, you need to use the labels informations. Python scatter plot with colors corresponding to strings should help you.