5 Best Ways to Draw Vertical Bar Plots with Nested Grouping by Two Categorical Variables in Seaborn

by

💡 Problem Formulation: When dealing with categorical data, it is often insightful to visualize the distribution across multiple group levels. This article explores methods of drawing vertical bar plots using Python’s Pandas and Seaborn libraries, focusing on nested grouping by two categorical variables. For instance, you have a dataset with ‘Brand’ and ‘Year’ as categories and you want to see the sales figures nested within each ‘Brand’, across different ‘Years’.

Method 1: Using Seaborn’s catplot()

This method leverages Seaborn’s catplot() function, which is a high-level interface for drawing categorical plots. It can handle nested grouping very effortlessly when the kind parameter is set to ‘bar’. You would specify your two categorical variables for the x and hue parameters respectively, and your numerical variable for the y parameter.

Here’s an example:

import seaborn as sns
import pandas as pd

# Example dataset
data = pd.DataFrame({
    'Brand': ['A', 'A', 'B', 'B'],
    'Year': [2019, 2020, 2019, 2020],
    'Sales': [200, 150, 300, 250]
})

# Create the nested bar plot
g = sns.catplot(x="Brand", hue="Year", y="Sales", data=data, kind="bar")
g.despine(left=True)
g.set_axis_labels("Brand", "Sales")

The output will be a Seaborn figure with vertical bars representing sales, nested by ‘Year’ within each ‘Brand’ category.

This code snippet first loads the necessary libraries (Seaborn and Pandas) and then defines a sample dataset. The dataset is then passed to the catplot() function with the appropriate variables mappings to create a nested bar plot. The despine() and set_axis_labels() functions are used for visual refinement.

Method 2: Utilizing FacetGrid() with a Bar Plot

FacetGrid() allows you to produce a grid of plots based on the values of one or more categorical variables. This can be particularly effective when you need to create separate plots for each level of a categorical variable, and on each of them, plot a bar chart with nested grouping by another variable.

Here’s an example:

import seaborn as sns
import matplotlib.pyplot as plt

# Define a custom function to draw a nested bar plot
def draw_bars(x, y, **kwargs):
    sns.barplot(x=x, y=y, **kwargs)

g = sns.FacetGrid(data, col="Brand")
g.map_dataframe(draw_bars, x="Year", y="Sales")
plt.show()

You will see multiple bar plots, each corresponding to a separate ‘Brand’, and within each plot, ‘Sales’ are shown for different ‘Years’.

In this example, FacetGrid() is used to create a grid layout, where each subplot corresponds to a different ‘Brand’. A custom function draw_bars() is used with map_dataframe() to draw individual bar plots within each subplot of the grid. This method offers increased customization for complex datasets.

Method 3: Using pointplot() for a Stacked Effective Presentation

Seaborn’s pointplot() function can create bar plots that represent the same data but emphasize continuity among the categorical variable. This approach could be an effective visualization technique to show trends more clearly.

Here’s an example:

import seaborn as sns

# Setting the Seaborn style
sns.set(style="whitegrid")

# Draw a pointplot to show nested grouping
g = sns.pointplot(x="Brand", y="Sales", hue="Year", data=data, palette="dark", markers=["o", "s"], linestyles=["-", "--"])
g.set_xlabel("Brand")
g.set_ylabel("Sales")
plt.show()

The output resembles a bar plot but with points connected by lines, emphasizing the trend for ‘Sales’ by ‘Year’ within each ‘Brand’ category.

The code demonstrates a variation in the presentation of grouped data by using pointplot(), where a dot represents the value of sales, and lines connect dots of the same ‘Year’. Different markers and line styles distinguish between the ‘Years’.

Method 4: Combining barplot() with dodge Parameter

Seaborn’s barplot() function can be used with the dodge parameter set to True, which serves to nest groups by creating side-by-side bars, helping in comparing two categorical variables more precisely.

Here’s an example:

import seaborn as sns

# Create the dodged barplot
sns.barplot(x="Brand", y="Sales", hue="Year", data=data, dodge=True)
plt.show()

The output will be a bar plot with sales nested by ‘Year’ shown side by side within each ‘Brand’ section.

The example code uses barplot() with the dodge=True parameter to create bars side by side for nested groups. This is straightforward and useful when comparing differences between groups directly.

Bonus One-Liner Method 5: Quick Nested Bar Plot with catplot()

A one-liner approach using Seaborn’s catplot() function that quickly generates a nested bar plot with minimal input.

Here’s an example:

sns.catplot('Brand', 'Sales', hue='Year', data=data, kind='bar', height=5, aspect=2).despine(left=True)

The simplified output produces a vertical bar plot which groups ‘Sales’ by both ‘Brand’ and ‘Year’ in a quick and efficient manner, ideal for rapid exploration of data.

This method employs a compact version of the previously mentioned catplot() function. The plot is configured with just a single line of code, which is ideal for quick data inspections where fine-tuned customization is not a primary concern.

Summary/Discussion

  • Method 1: catplot(). Highly customizable. Allows for additional plot types beyond bar plots. Can be verbose for quick plotting tasks.
  • Method 2: FacetGrid() with custom function. Great for comparisons within subsets. Requires more coding and setup compared to other methods.
  • Method 3: pointplot() for trend emphasis. Excellent for showing trends. Less immediately intuitive for representing specific value comparisons.
  • Method 4: barplot() with dodge parameter. Direct comparison of categorical variables. Limited to side-by-side bar presentation.
  • Method 5: One-liner catplot(). Efficient and fast. Less flexible for customization and may require additional coding for complex dataset structures.