19

I'm trying to combine a normal matplotlib.pyplot plt.plot(x,y) with variable y as a function of variable x with a boxplot. However, I only want a boxplot on certain (variable) locations of x but this does not seem to work in matplotlib?

Serenity
  • 35,289
  • 20
  • 120
  • 115
ruben baetens
  • 2,806
  • 6
  • 25
  • 31

2 Answers2

29

Are you wanting something like this? The positions kwarg to boxplot allows you to place the boxplots at arbitrary positions.

import matplotlib.pyplot as plt
import numpy as np

# Generate some data...
data = np.random.random((100, 5))
y = data.mean(axis=0)
x = np.random.random(y.size) * 10
x -= x.min()
x.sort()

# Plot a line between the means of each dataset
plt.plot(x, y, 'b-')

# Save the default tick positions, so we can reset them...
locs, labels = plt.xticks() 

plt.boxplot(data, positions=x, notch=True)

# Reset the xtick locations.
plt.xticks(locs)
plt.show()

enter image description here

Joe Kington
  • 275,208
  • 71
  • 604
  • 463
  • Yes, thank you, this is exactly what i wanted. Iwas always trying to do something like plt.plot([x,y]) in boxplot, but failed ... seemsi misunderstand the kwarg `positions` – ruben baetens May 10 '11 at 07:22
  • +1 for showing how to conserve the xtick locations. When working with axes the commands are `locs = ax.get_xticks()` and `ax.set_xticks(locs)` – MrCyclophil Mar 05 '16 at 20:01
  • Additionally to my comment above one actually needs to add `ax.set_xticklabels(locs)` as well to get the labels right. – MrCyclophil Mar 06 '16 at 18:48
  • Great answer without the need to use seaborn! Great! – seralouk May 07 '22 at 13:36
0

This is what has worked for me:

  1. plot box-plot
  2. get boxt-plot x-axis tick locations
  3. use box-plot x-axis tick locations as x-axis values for the line plot
# Plot Box-plot
ax.boxplot(data, positions=x, notch=True)
# Get box-plot x-tick locations
locs=ax.get_xticks()

# Plot a line between the means of each dataset
# x-values = box-plot x-tick locations
# y-values = means
ax.plot(locs, y, 'b-')
jumbofiatco
  • 21
  • 5