Statistical Visualization with Seaborn

# CHAPTER 19

Statistical Visualization with Seaborn #

1. Chapter Introduction #

2. Seaborn Basics and Installation #

Install it using !pip install seaborn.

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

# Load a built-in sample dataset from Seaborn
df = sns.load_dataset('tips')

print(df.head(2))
# Output:
#    total_bill   tip     sex smoker  day    time  size
# 0       16.99  1.01  Female     No  Sun  Dinner     2
# 1       10.34  1.66    Male     No  Sun  Dinner     3

3. Distribution Plots (Histograms with Density) #

Seaborn's histplot is an upgrade over Matplotlib's histogram. It can easily add a KDE (Kernel Density Estimate) line to show the smooth distribution of data.

# Create a histogram of Total Bills, add a KDE curve
sns.histplot(data=df, x='total_bill', kde=True, color='purple')

# We still use Matplotlib to add the title!
plt.title("Distribution of Total Bills")
plt.show()

4. Categorical Plots (Boxplots) #

If you want to compare numerical data across different categories, a Boxplot is the industry standard. It shows the median, the spread (quartiles), and identifies outliers (dots outside the lines).

# Compare total bills across different days of the week
# Notice how Seaborn takes the DataFrame directly via the 'data' parameter!
sns.boxplot(data=df, x='day', y='total_bill', palette='Set2')
plt.title("Total Bills by Day")
plt.show()

5. Regression Plots (Finding Trends) #

Instead of a basic scatter plot, Seaborn's lmplot (Linear Model Plot) draws a scatter plot *and* automatically calculates and draws a line of best fit through the data.

# Does a higher bill result in a higher tip?
sns.lmplot(data=df, x='total_bill', y='tip', scatter_kws={'alpha':0.5})
plt.title("Bill vs Tip (with Regression Line)")
plt.show()

6. The Pair Plot (The Magic Function) #

When you first load a dataset, you want to see how *every* numerical variable correlates with every other variable. Doing this manually takes hours. Seaborn's pairplot does it in one line.

# Draws scatter plots for every pair of variables, and histograms on the diagonal
sns.pairplot(df, hue='sex') # 'hue' color-codes the dots based on Gender!
plt.show()

7. The Correlation Heatmap #

A heatmap uses color to represent the strength of mathematical correlation between variables (from -1 to 1).

# 1. Calculate the correlation matrix (Pandas function)
# Note: We must select only numeric columns first!
numeric_df = df.select_dtypes(include=['float64', 'int64'])
corr_matrix = numeric_df.corr()

# 2. Plot the Heatmap (Seaborn function)
# annot=True puts the actual numbers inside the colored squares
sns.heatmap(corr_matrix, annot=True, cmap='coolwarm', vmin=-1, vmax=1)
plt.title("Correlation Heatmap")
plt.show()

8. Common Mistakes #

• Forgetting plt.show(): Even though you are using Seaborn functions (sns.), Seaborn relies on Matplotlib (plt) under the hood to actually render the image. You must still call plt.show().

• Passing text to a heatmap: df.corr() will crash if your DataFrame contains text columns (like "Name"). Always filter for numerical columns before running a correlation matrix.

9. MCQs #

10. Interview Questions #

• Q: Explain what a Correlation Heatmap is. How do you interpret a value of 0.9 versus a value of -0.9?

• Q: If you receive a dataset with 15 numerical columns and need to quickly find which variables are highly correlated, what Seaborn function would you use?

11. Summary #

12. Next Chapter Recommendation #