How to set seed for jitter in seaborn stripplot?

Question

I am trying to reproduce stripplots exactly so that I can draw lines and write on them reliably. However, when I produce a stripplot with jitter the jitter is random and prevents me from achieving my goal.

I have blindly tried some rcParams I found in other Stack Overflow posts, such as mpl.rcParams['svg.hashsalt'] which hasn't worked. I also tried setting a seed for random.seed() without success.

The code I am running looks like the following.

import seaborn as sns
import matplotlib.pyplot as plt
import random

plt.figure(figsize=(14,9))

random.seed(123)

catagories = []
values = []

for i in range(0,200):
    n = random.randint(1,3)
    catagories.append(n)

for i in range(0,200):
    n = random.randint(1,100)
    values.append(n)

sns.stripplot(catagories, values, size=5)
plt.title('Random Jitter')
plt.xticks([0,1,2],[1,2,3])
plt.show()

This code generates a stripplot just like I want. But if you run the code twice you will get different placements for the points, due to the jitter. The graph I am making requires jitter to not look ridiculous, but I want to write on the graph. However there is no way to know the exact positions of the points before running the code, which then changes every time the code is run.

Is there any way to set a seed for the jitter in seaborn stripplots to make them perfectly reproduceable?

Answer 1

• jitter is determined by scipy.stats.uniform
• uniform is class uniform_gen(scipy.stats._distn_infrastructure.rv_continuous)
• Which is a subclass of class rv_continuous(rv_generic)
• Which has a seed parameter, and uses np.random
• Therefore, use np.random.seed()
  • It needs to be called before each plot. In the case of the example, np.random.seed(123) must be inside the loop.

from the Stripplot docstring

jitter : float, ``True``/``1`` is special-cased, optional
    Amount of jitter (only along the categorical axis) to apply. This
    can be useful when you have many points and they overlap, so that
    it is easier to see the distribution. You can specify the amount
    of jitter (half the width of the uniform random variable support),
    or just use ``True`` for a good default.

From class _StripPlotter in categorical.py

• jitter is calculated with scipy.stats.uniform

from scipy import stats

class _StripPlotter(_CategoricalScatterPlotter):
    """1-d scatterplot with categorical organization."""
    def __init__(self, x, y, hue, data, order, hue_order,
                 jitter, dodge, orient, color, palette):
        """Initialize the plotter."""
        self.establish_variables(x, y, hue, data, orient, order, hue_order)
        self.establish_colors(color, palette, 1)

        # Set object attributes
        self.dodge = dodge
        self.width = .8

        if jitter == 1:  # Use a good default for `jitter = True`
            jlim = 0.1
        else:
            jlim = float(jitter)
        if self.hue_names is not None and dodge:
            jlim /= len(self.hue_names)
        self.jitterer = stats.uniform(-jlim, jlim * 2).rvs

from the rv_continuous docstring

    seed : {None, int, `~np.random.RandomState`, `~np.random.Generator`}, optional
        This parameter defines the object to use for drawing random variates.
        If `seed` is `None` the `~np.random.RandomState` singleton is used.
        If `seed` is an int, a new ``RandomState`` instance is used, seeded
        with seed.
        If `seed` is already a ``RandomState`` or ``Generator`` instance,
        then that object is used.
        Default is None.

Using your code with np.random.seed

• All the plot points are the same

import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np

fig, axes = plt.subplots(2, 3, figsize=(12, 12))
for x in range(6):

    np.random.seed(123)

    catagories = []
    values = []

    for i in range(0,200):
        n = np.random.randint(1,3)
        catagories.append(n)

    for i in range(0,200):
        n = np.random.randint(1,100)
        values.append(n)

    row = x // 3
    col = x % 3
    axcurr = axes[row, col]

    sns.stripplot(catagories, values, size=5, ax=axcurr)
    axcurr.set_title(f'np.random jitter {x+1}')
plt.show()

using just random

• The plot points move around

import seaborn as sns
import matplotlib.pyplot as plt
import random

fig, axes = plt.subplots(2, 3, figsize=(12, 12))
for x in range(6):

    random.seed(123)

    catagories = []
    values = []

    for i in range(0,200):
        n = random.randint(1,3)
        catagories.append(n)

    for i in range(0,200):
        n = random.randint(1,100)
        values.append(n)

    row = x // 3
    col = x % 3
    axcurr = axes[row, col]

    sns.stripplot(catagories, values, size=5, ax=axcurr)
    axcurr.set_title(f'random jitter {x+1}')
plt.show()

Using random for the data and np.random.seed for the plot

fig, axes = plt.subplots(2, 3, figsize=(12, 12))
for x in range(6):

    random.seed(123)

    catagories = []
    values = []

    for i in range(0,200):
        n = random.randint(1,3)
        catagories.append(n)

    for i in range(0,200):
        n = random.randint(1,100)
        values.append(n)

    row = x // 3
    col = x % 3
    axcurr = axes[row, col]

    np.random.seed(123)
    sns.stripplot(catagories, values, size=5, ax=axcurr)
    axcurr.set_title(f'np.random jitter {x+1}')
plt.show()