Control scatter plot y axis order in matplotlib

Question

I'm trying to control the y axis order on a matplotlib scatter plot but the ordering of the x and y axes in the data I have is causing the plot to be displayed incorrectly.

Here's some code to illustrate the problem and one sub-optimal attempt to make a solution.

import pandas as pd
from numpy import random
import matplotlib.pyplot as plt

# make some fake data
axes = ['a', 'b', 'c', 'd']
pairs = pd.DataFrame([(x, y) for x in axes for y in axes], columns=['x', 'y'])
pairs['value'] = random.randint(100, size=16) + 100
# remove the diagonal
pairs_nodiag = pairs[pairs['x'] != pairs['y']]
# zero the values for the diagonal
pairs_diag = pairs.copy()
pairs_diag.loc[pairs_diag['x'] == pairs_diag['y'], 'value'] = 0

fig, ax = plt.subplots(nrows=1, ncols=3, figsize=(5, 3))
scatter = ax[0].scatter(x=pairs['x'], y=pairs['y'], s=pairs['value'])
scatter = ax[1].scatter(x=pairs_nodiag['x'], y=pairs_nodiag['y'], s=pairs_nodiag['value'])
scatter = ax[2].scatter(x=pairs_diag['x'], y=pairs_diag['y'], s=pairs_diag['value'])

plt.show()

The left most is the raw data. The middle is the plot with the problem; I want the y axis to be the same as the left most plot. The right most plot is what I am after using a sub-optimal workaround. I'm sure there is a way of controlling the ordering on the axes but I'm not expert enough in Python yet to know exactly how to do this.

I think your workaround is not a workaround but it's the right way to do it. With the boolean indexing you get ('a', 'b') as a first value correctly, but of course this screws up the order. — Francesco
– Francesco, Commented Jul 6, 2020 at 11:57
I suppose it's a valid work around but in reality, the data I get isn't complete so it would be tiresome to have to patch it up to ensure the plotting works. — Andrew Chisholm
– Andrew Chisholm, Commented Jul 6, 2020 at 13:07
Unfortunately I think you have to keep some placeholder for values that you don't want to plot. I would use None instead of 0 — Francesco
– Francesco, Commented Jul 6, 2020 at 14:20

Stef · Accepted Answer · 2020-07-06 15:21:13Z

3

You need to create your own StringCategoryConverter with your desired mapping (matplotlib by default maps strings to numbers in the sequence the occur).

import matplotlib.category as mcat

# insert the following before scatter = ax[1].scatter(...
units = mcat.UnitData(sorted(pairs_nodiag.y.unique()))
ax[1].yaxis.set_units(units)
ax[1].yaxis.set_major_locator(mcat.StrCategoryLocator(units._mapping))
ax[1].yaxis.set_major_formatter(mcat.StrCategoryFormatter(units._mapping))

UPDATE: The following is the official way to do it without using _mapping:

import matplotlib

# insert the following before scatter = ax[1].scatter(...
scc = matplotlib.category.StrCategoryConverter()
units = scc.default_units(sorted(pairs_nodiag.y.unique()), ax[1].yaxis)
axisinfo = scc.axisinfo(units, ax[1].yaxis)
ax[1].yaxis.set_major_locator(axisinfo.majloc)
ax[1].yaxis.set_major_formatter(axisinfo.majfmt)

edited Jul 6, 2020 at 15:21

answered Jul 6, 2020 at 14:57

Stef

30.9k3 gold badges34 silver badges60 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Nilesh Ingle Over a year ago

@Stef: Thanks. Learned new today matplotlib.category to sort axis. Upvoted! Deleted my answer as it not relevant anymore.

Collectives™ on Stack Overflow

Control scatter plot y axis order in matplotlib

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related