Seaborn — Statistical Charts in Python | Data Analyst Interview
🎨 Statistical Visualisation

Seaborn — Beautiful Statistical Charts in Python

Heatmaps, pairplots, violin plots. Built on Matplotlib, Seaborn turns one-line code into presentation-ready charts that interviewers love.

Book Mock Interview ← Back to Python
Core Concepts
What interviewers test in Seaborn

Seaborn turns one-line code into statistical charts that Matplotlib makes painful. Master these six and you’ll handle any EDA round.

1Distribution Plots

histplot, kdeplot, displot — understand the shape of one variable at a time. The first move in any EDA session.

  • histplot with kde=True
  • kdeplot for smooth density
  • Multiple distributions on one axis

2Categorical Plots

boxplot, violinplot, stripplot — compare distributions across categories. A staple of A/B test and product analytics interviews.

  • boxplot for outlier detection
  • violinplot for full distribution
  • countplot for category sizes

3Relational Plots

scatterplot, lineplot — visualise relationships between two numerical variables, optionally split by a third.

  • scatterplot with hue and size
  • lineplot with confidence bands
  • relplot for faceted grids

4Heatmaps & Correlation

heatmap — the easiest way to spot correlated features. Essential in any ML or feature-engineering interview round.

  • sns.heatmap(df.corr())
  • annot=True to show values
  • cmap=’coolwarm’ diverging scale

5Pairplot & FacetGrid

pairplot for all-vs-all relationships; FacetGrid for splitting any chart by a categorical column.

  • sns.pairplot(df, hue=’label’)
  • FacetGrid(col, row, hue)
  • catplot kind=’box’ or ‘bar’

6Themes & Palettes

set_theme, color_palette — clean styling without Matplotlib boilerplate. Saves critical time in live coding tests.

  • sns.set_theme(‘whitegrid’)
  • palette=’viridis’ or ‘Set2’
  • context=’talk’ for presentations
Interview Example 1
Correlation heatmap for feature selection

A staple of EDA rounds. Spot which features correlate with the target — and which are redundant.

import seaborn as sns
import matplotlib.pyplot as plt
 
sns.set_theme(style=‘whitegrid’)
fig, ax = plt.subplots(figsize=(10, 8))
 
sns.heatmap(df.corr(numeric_only=True),
  annot=True, fmt=‘.2f’, cmap=‘coolwarm’,
  center=0, square=True, ax=ax)
 
ax.set_title(‘Feature Correlation Matrix’, fontsize=14)
plt.tight_layout(); plt.show()
Interview Example 2
Box plot comparing two dimensions

“Compare distributions across two dimensions” — the hue parameter is the key move here.

fig, ax = plt.subplots(figsize=(10, 5))
 
sns.boxplot(data=df, x=‘category’, y=‘amount’,
  hue=‘is_member’, palette=‘Set2’, ax=ax)
 
ax.set_title(‘Order amount by category and membership’)
ax.set_ylabel(‘Amount (£)’)
plt.tight_layout(); plt.show()
Interview Questions
Real Seaborn questions asked in 2026

Seaborn questions are usually paired with statistical reasoning — not just “make this chart”.

When would you use Seaborn instead of Matplotlib?
Seaborn for statistical charts and quick EDA: pairplots, heatmaps, violin plots, and anything where sensible defaults matter. Matplotlib for fine-grained control, non-standard layouts, or production figures requiring precise customisation. In practice both coexist — Seaborn for the chart, Matplotlib for the polish.
What is the difference between a boxplot and a violin plot?
Boxplots show the five-number summary: median, quartiles, and outliers. Violin plots additionally show the full distribution shape via mirrored KDE curves. Use violin plots when distribution shape — such as bimodality — matters; boxplots when you only need location and spread.
What does the hue parameter do?
hue splits the data by a categorical variable and maps each level to a different colour. It turns a 2D plot into a 3-dimensional comparison without needing separate panels. Example: sns.scatterplot(x=’age’, y=’salary’, hue=’department’) shows all three relationships in one chart.
What is a pairplot and when do you use it?
pairplot shows scatter plots for every pair of numerical columns with histograms on the diagonal. It is the fastest way to spot correlations, clusters, and outliers across many features at once. Add hue to colour-code by a target class — extremely useful in the EDA phase of a classification problem.
When is a heatmap the wrong choice?
Heatmaps fail when the matrix is sparse (most cells empty), when values span many orders of magnitude without log scaling, or when you have too many rows/columns for the annotations to be readable. In those cases, consider a sorted bar chart or a hierarchically clustered heatmap instead.
Quick Reference
Seaborn cheat sheet

The functions that handle 90% of EDA tasks in analyst interviews.

FunctionPurpose
sns.set_theme(style)Apply global style — whitegrid, ticks, dark
sns.histplot(data, x, kde)Histogram with optional density curve
sns.boxplot(data, x, y, hue)Distribution comparison with outliers
sns.violinplot(data, x, y)Full distribution shape comparison
sns.scatterplot(data, x, y, hue)Two numerics plus a category
sns.lineplot(data, x, y)Time series with confidence bands
sns.heatmap(matrix, annot, cmap)Coloured grid for correlation matrices
sns.pairplot(df, hue)All-pairs scatter matrix
sns.countplot(data, x)Bar chart of category frequencies
sns.regplot(data, x, y)Scatter with regression line
sns.catplot(kind=’bar’)Categorical charts with faceting
sns.color_palette(‘Set2’)Apply a custom colour scheme

Ready to ace your EDA round?

Practise Seaborn charting questions with feedback on chart choice and statistical reasoning.

Book Free Seaborn Session
Subscribe
Subscribe