How to center align the plot over the xticks when offset by hue

Question

My boxplot seem not align with the x-tick of the plot. How to make the boxplot align with the x-tick?

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

df = pd.DataFrame([['0', 0.3],['1', 0.5],['2', 0.9],
                   ['0', 0.8],['1', 0.3],['2', 0.4],
                   ['0', 0.4],['1', 0.0],['2', 0.7]])

df.columns = ['label', 'score']

label_list = ['0', '1', '2']

fig = plt.figure(figsize=(8, 5))
g=sns.boxplot(x='label', y='score', data=df, hue='label', hue_order=label_list)
g.legend_.remove()

plt.show()

Answer 1

• As stated by ImportanceOfBeingErnest, remove hue='label', hue_order=label_list. That is responsible for shifting the boxes.
  • Applicable to sns.catplot and associated axes-level categorical plots.
• hue should be used to encode a different set of categories, not to encode the same category, 'label', multiple times.
  • The 'label' category is already encoded on the x-axis. Likewise, there doesn't need to be a legend, because the information is already encoded on the x-axis.
• Following are a few citations outlining graphical perception and color usage.
  • While some packages, like ggplot, and seaborn, may encode every category on the x-axis with a color, it's better to avoid unnecessary usage of color.
  • From Practical Rules for Using Color in Charts by Stephen Few, Perceptual Edge, in the Use Color Meaningfully and with Restraint section: the color usage in this plot, means nothing, and should be avoided.
    • Use color only when needed to serve a particular communication goal.
    • Use different colors only when they correspond to differences of meaning in the data.
  • As a general reference, see Graphical Perception and Graphical Methods for Analyzing Scientific Data, William S. Cleveland and Robert McGill, New Series, Vol. 229, No. 4716 (Aug. 30, 1985), pp. 828-833 (6 pages), published by: American Association for the Advancement of Science
    • Virginia Tech

Default Plot Without hue

ax = sns.boxplot(data=df, x='label', y='score')
_ = ax.set(title='Default Plot\nUnnecessary Use of Color')

Plot With a Single Color

ax = sns.boxplot(data=df, x='label', y='score', color='tab:blue')
_ = ax.set(title='Avoid Unnecessary Use of Color')

---

• If the order of categories on the x-axis is important, and not correctly ordered, one of the following options can be implemented:
  1. Specify the order with the order parameter.
     • order=['0', '1', '2'] or order=label_list
  2. Convert the df column to a category Dtype with pd.Categorical
     • df.label = pd.Categorical(values=df.label, categories=label_list, ordered=True)

Answer 2

You can add dodge=False into your boxplot line & that should fix this. That updated code would be as below

g=sns.boxplot(x='label', y='score', data=df, hue='label', hue_order=label_list, dodge=False)

You can then play with width to control the width (default width is 0.8) of box plots like below

g=sns.boxplot(x='label', y='score', data=df, hue='label', hue_order=label_list, dodge=False, width=.2)