How to Visualize Categorical Data with Pie Charts and Bar Plots in EDA

Visualizing categorical data is a crucial step in exploratory data analysis (EDA) as it helps reveal underlying patterns, trends, and insights that drive further analysis or decision-making. Two of the most widely used visualization techniques for categorical data are pie charts and bar plots. While both are designed for representing frequency distributions of categories, they serve slightly different purposes and have their own advantages. This article delves into how to effectively use pie charts and bar plots for visualizing categorical data during EDA, covering the scenarios, advantages, limitations, and best practices for each.

Understanding Categorical Data

Categorical data refers to variables that represent categories or groups, such as:

• Nominal data: unordered categories (e.g., colors, brands, types of animals)

• Ordinal data: ordered categories (e.g., education levels, satisfaction ratings)

Such data cannot be quantified on a numerical scale but can be grouped and counted. Therefore, visualization techniques must convey the frequency or proportion of these groups.

Importance of Visualizing Categorical Data in EDA

• Identifies dominant categories

• Detects data imbalances

• Highlights potential outliers or rare classes

• Provides intuitive visual summaries for stakeholders

• Guides feature engineering and model selection

Pie Charts: Proportional Representation

What is a Pie Chart?

A pie chart is a circular graph divided into slices, where each slice represents a category’s contribution to the whole. The size of each slice is proportional to the frequency or percentage of the category it represents.

When to Use Pie Charts

• When displaying the part-to-whole relationship

• For datasets with a small number of categories (typically less than 5-6)

• To compare relative proportions rather than exact values

How to Create a Pie Chart in Python (Matplotlib)

python
CopyEdit
import matplotlib.pyplot as plt

# Sample data
categories = ['A', 'B', 'C', 'D']
values = [45, 25, 15, 15]

# Create pie chart
plt.figure(figsize=(6,6))
plt.pie(values, labels=categories, autopct='%1.1f%%', startangle=140)
plt.title('Category Distribution')
plt.show()

Pros of Pie Charts

• Simple and intuitive

• Emphasizes proportions visually

• Great for presentations and non-technical audiences

Cons of Pie Charts

• Difficult to compare similar-sized slices

• Not ideal for many categories

• Misleading if not scaled or ordered correctly

Best Practices for Pie Charts

• Keep categories to a minimum (ideally under six)

• Display percentages or labels directly on the chart

• Sort categories to aid visual clarity (e.g., largest to smallest)

• Use contrasting colors for different slices

Bar Plots: Frequency and Comparison

What is a Bar Plot?

A bar plot (or bar chart) represents categorical data with rectangular bars. Each bar’s length corresponds to the category’s frequency or value. Bar plots can be vertical or horizontal.

When to Use Bar Plots

• When the number of categories is larger

• For comparing absolute values across categories

• When data has clear ordering (ordinal variables)

• To highlight differences between categories

How to Create a Bar Plot in Python (Seaborn)

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

# Sample data
data = pd.DataFrame({
    'Category': ['A', 'B', 'C', 'D', 'E'],
    'Count': [50, 30, 20, 10, 5]
})

# Create bar plot
plt.figure(figsize=(8,6))
sns.barplot(x='Category', y='Count', data=data)
plt.title('Category Frequency')
plt.show()

Pros of Bar Plots

• Easy to interpret and compare

• Suitable for large numbers of categories

• Flexible for both nominal and ordinal data

• Can include error bars, annotations, and colors

Cons of Bar Plots

• Less visually appealing than pie charts

• Can be cluttered with too many categories

• Might require axis scaling and sorting

Best Practices for Bar Plots

• Sort bars by frequency or relevance

• Use color coding for subgroups if applicable

• Label axes clearly

• Use horizontal orientation for long category names

• Use grouped or stacked bars for multivariate analysis

Pie Charts vs Bar Plots: Comparison Table

Common Use Cases in EDA

1. Gender Distribution

A dataset with “Male” and “Female” categories can be visualized using a pie chart to show the overall distribution.

python
CopyEdit
labels = ['Male', 'Female']
sizes = [60, 40]
plt.pie(sizes, labels=labels, autopct='%1.1f%%')
plt.title('Gender Distribution')
plt.show()

2. Product Category Frequencies

A bar plot can show how many products fall into each category such as electronics, furniture, clothing, etc.

python
CopyEdit
sns.countplot(data=dataset, x='Product_Category')
plt.xticks(rotation=45)
plt.title('Product Category Frequency')
plt.show()

3. Customer Satisfaction Ratings

If categories are “Very Unsatisfied”, “Unsatisfied”, “Neutral”, “Satisfied”, “Very Satisfied”, a bar plot is better suited to show the trend and skewness in responses.

Combining Pie Charts and Bar Plots

In some EDA reports, it’s beneficial to combine both visualizations for the same variable:

• Use a pie chart to show proportions

• Use a bar chart to emphasize differences in actual counts

This dual-approach ensures that the audience grasps both aspects of the data — the relative distribution and the actual magnitude.

Tips for Effective Categorical Data Visualization

• Always include clear labels and legends

• Use consistent color schemes across visualizations

• Avoid using pie charts with too many or similar-sized categories

• Prefer bar plots for statistical analysis and pie charts for illustrative summaries

• Check for zero or missing values before plotting

• For ordinal data, ensure categories are sorted logically

Conclusion

Pie charts and bar plots are indispensable tools for visualizing categorical data during exploratory data analysis. While pie charts are more suitable for highlighting proportions in small, well-defined groups, bar plots excel in showing frequency comparisons, especially when dealing with larger or more complex datasets. Using these tools effectively can uncover meaningful patterns in your data, aiding better decisions, communication, and insights in the EDA process. Always choose the appropriate visualization based on your data characteristics, analysis goals, and the audience for your results.