⬡ Hub
Skip to content

Seaborn: Relational Plots

Relational plots in Seaborn are designed to visualize the relationships between two or more quantitative variables. The relplot() function is the primary entry point for creating these plots, and it can generate either scatter plots or line plots, with the ability to show multiple subsets of the data using a faceted grid.

1. Scatter Plots (scatterplot and relplot(kind='scatter'))

Scatter plots are ideal for visualizing the relationship between two continuous variables. Each point represents an observation, and its position on the x and y axes corresponds to the values of the two variables.

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

# Load a built-in dataset
tips = sns.load_dataset('tips')

plt.figure(figsize=(8, 6))
sns.scatterplot(data=tips, x='total_bill', y='tip', hue='smoker', style='time', size='size', palette='viridis', sizes=(20, 200), alpha=0.7)
plt.title('Total Bill vs Tip (by Smoker and Time of Day)')
plt.xlabel('Total Bill ($)')
plt.ylabel('Tip ($)')
plt.legend(title='Features', bbox_to_anchor=(1.05, 1), loc='upper left') # Place legend outside
plt.tight_layout()
plt.show()

# Using relplot for a scatter plot with faceting
sns.relplot(data=tips, x='total_bill', y='tip', col='time', row='smoker', hue='day', kind='scatter', height=3, aspect=1.2, palette='Set1')
plt.suptitle('Total Bill vs Tip by Time, Smoker, and Day', y=1.02)
plt.show()

2. Line Plots (lineplot and relplot(kind='line'))

Line plots are best suited for visualizing the relationship between two quantitative variables where one of the variables (usually on the x-axis) represents a time series or an ordered sequence. Seaborn's lineplot automatically aggregates data points within bins and plots the mean and a confidence interval around it.

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

# Load a built-in time-series dataset
fmri = sns.load_dataset('fmri')

plt.figure(figsize=(10, 6))
sns.lineplot(data=fmri, x='timepoint', y='signal', hue='event', style='region', markers=True, dashes=False)
plt.title('fMRI Signal over Time (by Event and Region)')
plt.xlabel('Timepoint')
plt.ylabel('Signal')
plt.legend(title='Features', bbox_to_anchor=(1.05, 1), loc='upper left')
plt.tight_layout()
plt.show()

# Using relplot for a line plot with faceting
sns.relplot(data=fmri, x='timepoint', y='signal', col='region', hue='event', kind='line', height=3, aspect=1.2, col_wrap=2)
plt.suptitle('fMRI Signal by Timepoint, Region, and Event', y=1.02)
plt.show()

3. Customizing Relational Plots

Both scatterplot and lineplot (and thus relplot) offer extensive customization options through various parameters:

  • hue: Categorical variable to map plot aspects to different colors.
  • size: Quantitative variable to map plot aspects to different sizes.
  • style: Categorical variable to map plot aspects to different marker styles (for scatter) or line styles (for line).
  • palette: Colormap or list of colors.
  • alpha: Transparency.
  • markers: Boolean, whether to draw markers on the line plot.
  • dashes: Boolean or sequence of dashes for line plot.
  • x_ci, y_ci: Confidence interval size (for line plots).
  • err_style: Style of error bars (e.g., 'band', 'bars').

Further Topics:

  • jointplot for combined relational and distribution plots.
  • pairplot for visualizing pairwise relationships across an entire DataFrame.
  • Regression lines within scatter plots using lmplot.
  • Handling large datasets in relational plots.

Relational plots are invaluable for exploring the interplay between variables in your dataset, helping you uncover correlations, trends, and group-specific behaviors. Seaborn's high-level interface makes these complex visualizations surprisingly easy to generate.