Themes and Final Polish

Apply professional themes to visualizations. Fine-tune every aspect of your plots. Create publication-ready figures. Learn best practices for presenting research visualizations.

Learning Objectives

Apply seaborn themes with sns.set_theme()
Customize plot aesthetics with .theme()
Control figure size and resolution
Add annotations and text to plots
Export publication-quality figures
Follow best practices for research visualizations

Key Questions

How do I apply professional themes to my plots?
How do I fine-tune visual elements?
How do I add annotations and notes to plots?
How do I save high-quality figures for publication?
What makes a visualization publication-ready?

The Importance of Visual Polish

You’ve learned to create powerful visualizations with seaborn.objects. Now let’s make them publication-ready:

Professional appearance increases credibility
Consistent styling across figures looks polished
Proper formatting helps reviewers and stakeholders take your work seriously
Attention to detail reflects the quality of your research

Setting Up

import seaborn as sns
import seaborn.objects as so
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Load data
penguins = sns.load_dataset("penguins").dropna()

Seaborn Themes

Seaborn provides several pre-built themes that control the overall appearance:

# Default seaborn theme
sns.set_theme()

(
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g", color="species")
    .add(so.Dot())
    .label(title="Default Theme")
)

Available Themes

themes = ["darkgrid", "whitegrid", "dark", "white", "ticks"]

for theme in themes:
    sns.set_theme(style=theme)
    (
        so.Plot(penguins, x="flipper_length_mm", y="body_mass_g", color="species")
        .add(so.Dot(pointsize=8))
        .label(title=f"Theme: {theme}")
    )

Theme descriptions:

darkgrid - Gray background with white grid lines (good for presentations)
whitegrid - White background with gray grid lines (clean, professional)
dark - Dark background without grid (minimalist)
white - White background without grid (publication standard)
ticks - White background with tick marks (classic academic)

Setting Context

Control the scale of plot elements for different outputs:

# For papers (smallest)
sns.set_theme(context="paper")

# For notebooks (medium, default)
sns.set_theme(context="notebook")

# For talks (larger)
sns.set_theme(context="talk")

# For posters (largest)
sns.set_theme(context="poster")

Combining Style and Context

# Perfect for academic papers
sns.set_theme(style="white", context="paper")

# Great for presentations
sns.set_theme(style="darkgrid", context="talk")

# Ideal for posters
sns.set_theme(style="whitegrid", context="poster")

Fine-Tuning with .theme()

For plot-specific customization, use .theme():

(
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g", color="species")
    .add(so.Dot(pointsize=8, alpha=0.7))
    .theme({
        "axes.facecolor": "#f8f9fa",     # Light gray background
        "axes.edgecolor": "#343a40",      # Dark edges
        "axes.linewidth": 1.5,            # Thicker borders
        "grid.color": "white",            # White grid lines
        "grid.linewidth": 1.2,            # Thicker grid
    })
    .label(
        title="Custom Theme Example",
        x="Flipper Length (mm)",
        y="Body Mass (g)",
        color="Species"
    )
)

Common Theme Parameters

Parameter	Controls	Example Values
`axes.facecolor`	Background color	`"white"`, `"#f0f0f0"`
`axes.edgecolor`	Border color	`"black"`, `"gray"`
`axes.linewidth`	Border thickness	`1.0`, `2.0`
`grid.color`	Grid line color	`"gray"`, `"#e0e0e0"`
`grid.linewidth`	Grid thickness	`0.5`, `1.0`
`font.size`	Base font size	`10`, `12`, `14`
`axes.labelsize`	Axis label size	`"medium"`, `12`
`axes.titlesize`	Title size	`"large"`, `14`

Controlling Figure Size

Figure size is crucial for different outputs:

# Small for inline display
(
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g")
    .add(so.Dot())
    .theme({"figure.figsize": (6, 4)})
)

# Large for presentations
(
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g")
    .add(so.Dot())
    .theme({"figure.figsize": (12, 8)})
)

# Wide for multi-panel
(
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g")
    .facet(col="species")
    .add(so.Dot())
    .theme({"figure.figsize": (15, 4)})
)

Complete Research Example

Let’s create a publication-ready figure for a research paper:

# Set publication theme
sns.set_theme(style="white", context="paper")

# Create realistic program data
np.random.seed(42)
time_points = [0, 3, 6, 9, 12]
groups = ['Control', 'Treatment']

program_data = []
for group in groups:
    for time in time_points:
        n = 50
        if group == 'Control':
            mean = 50 + time * 0.4 + np.random.normal(0, 1)
        else:
            mean = 50 + time * 1.2 + np.random.normal(0, 1)

        values = np.random.normal(mean, 8, n)
        for val in values:
            program_data.append({
                'group': group,
                'month': time,
                'outcome': val
            })

program_df = pd.DataFrame(program_data)

# Create publication-quality plot
plot = (
    so.Plot(program_df, x="month", y="outcome", color="group")
    # Data layer
    .add(so.Dot(alpha=0.2, pointsize=4))
    # Statistical layers
    .add(so.Line(linewidth=3), so.Agg())
    .add(so.Band(alpha=0.15), so.Est())
    # Styling
    .scale(
        color=so.Nominal(["#666666", "#E69F00"]),  # Colorblind-safe
        x=so.Continuous().tick(at=time_points),
        y=(30, 80)
    )
    .label(
        title="Impact of Microfinance Training on Household Savings\nRandomized Controlled Trial, 2022-2023",
        x="Months Since Baseline",
        y="Monthly Savings (KSh, thousands)",
        color="Group"
    )
    .theme({
        "axes.facecolor": "white",
        "axes.edgecolor": "#333333",
        "axes.linewidth": 1.2,
        "grid.color": "#e5e5e5",
        "grid.linewidth": 0.8,
        "figure.figsize": (8, 5),
        "axes.labelsize": 11,
        "axes.titlesize": 12,
        "legend.fontsize": 10
    })
)

# Display
plot

Saving Figures

Basic Saving

# Save as PNG (good for most uses)
plot.save("figure1_program_impact.png", dpi=300)

# Save as PDF (best for publications)
plot.save("figure1_program_impact.pdf")

# Save as SVG (scalable, good for editing)
plot.save("figure1_program_impact.svg")

Resolution Settings

# Lower resolution (presentations, web)
plot.save("figure_web.png", dpi=150)

# Medium resolution (reports)
plot.save("figure_report.png", dpi=300)

# High resolution (print publications)
plot.save("figure_print.png", dpi=600)

Controlling Output Size

# Specific dimensions in inches
plot = (
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g")
    .add(so.Dot())
    .theme({"figure.figsize": (7, 5)})  # Width=7", Height=5"
)

plot.save("figure_sized.png", dpi=300)
# Results in: 2100 x 1500 pixel image (7*300 x 5*300)

Adding Text and Annotations

While seaborn.objects doesn’t have built-in annotation, we can access the underlying matplotlib axes:

# Create plot
plot = (
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g", color="species")
    .add(so.Dot(alpha=0.6))
    .label(
        title="Penguin Measurements with Annotation",
        x="Flipper Length (mm)",
        y="Body Mass (g)",
        color="Species"
    )
)

# Get the matplotlib figure and axes
fig = plot.plot()._figure
ax = fig.axes[0]

# Add annotation
ax.annotate(
    'Largest penguins',
    xy=(230, 6000),
    xytext=(215, 5500),
    arrowprops=dict(arrowstyle='->', color='red', lw=1.5),
    fontsize=10,
    color='red'
)

# Add text box
ax.text(
    175, 3200,
    'Smallest penguins\nfound here',
    fontsize=9,
    bbox=dict(boxstyle='round', facecolor='wheat', alpha=0.5)
)

plt.tight_layout()
plt.show()

Multi-Panel Publication Figure

Create a complete figure with multiple subplots:

# Set theme
sns.set_theme(style="white", context="paper")

# Panel A: Scatter with regression
panel_a = (
    so.Plot(penguins, x="flipper_length_mm", y="body_mass_g", color="species")
    .add(so.Dot(alpha=0.5, pointsize=5))
    .add(so.Line(linewidth=2), so.PolyFit(order=1))
    .scale(color="colorblind")
    .label(
        title="A. Flipper Length vs Body Mass",
        x="Flipper Length (mm)",
        y="Body Mass (g)",
        color="Species"
    )
)

# Panel B: Distributions
panel_b = (
    so.Plot(penguins, x="body_mass_g")
    .facet(row="species")
    .add(so.Bars(), so.Hist(bins=20))
    .label(
        title="B. Body Mass Distributions",
        x="Body Mass (g)",
        y="Count"
    )
)

# Panel C: Means with error bars
panel_c = (
    so.Plot(penguins, x="species", y="body_mass_g", color="species")
    .add(so.Dot(pointsize=10), so.Agg())
    .add(so.Dash(width=0.5), so.Est())
    .scale(color="colorblind")
    .label(
        title="C. Mean Body Mass by Species",
        x="Species",
        y="Body Mass (g)",
        color="Species"
    )
)

# Display each panel
panel_a
panel_b
panel_c

Style Guide for Research Visualizations

For Academic Papers

sns.set_theme(style="white", context="paper")
# - Clean, minimal design
# - Black and white friendly (use shapes)
# - 300+ dpi resolution
# - PDF or TIFF format

For Presentations

sns.set_theme(style="darkgrid", context="talk")
# - High contrast
# - Large text and points
# - Bold colors
# - 150 dpi is sufficient
# - PNG format

For Policy Briefs

sns.set_theme(style="whitegrid", context="notebook")
# - Clear, clean design
# - Color for emphasis
# - Simple, direct message
# - 300 dpi
# - PNG or PDF

For Posters

sns.set_theme(style="whitegrid", context="poster")
# - Extra large text
# - Bold colors and markers
# - Visible from distance
# - High resolution (300+ dpi)
# - PDF format

Exercises

Exercise 1: Theme Comparison

Create the same plot with three different themes:

One appropriate for an academic paper
One for a presentation
One for a poster

Compare them - what differences do you notice?

# Your code here

Solution 1

# Academic paper
sns.set_theme(style="white", context="paper")
(
    so.Plot(penguins, x="bill_length_mm", y="bill_depth_mm", color="species")
    .add(so.Dot(pointsize=6), marker="species")
    .scale(color="colorblind")
    .label(title="For Academic Paper")
)

# Presentation
sns.set_theme(style="darkgrid", context="talk")
(
    so.Plot(penguins, x="bill_length_mm", y="bill_depth_mm", color="species")
    .add(so.Dot(pointsize=12))
    .label(title="For Presentation")
)

# Poster
sns.set_theme(style="whitegrid", context="poster")
(
    so.Plot(penguins, x="bill_length_mm", y="bill_depth_mm", color="species")
    .add(so.Dot(pointsize=18))
    .label(title="For Poster")
)

Notice:

Font sizes increase from paper → talk → poster
Paper version uses both color and shape (for B&W printing)
Talk version has grid for easier reading at distance
Poster version has largest points for visibility

Exercise 2: Publication-Ready Figure

Create a complete, publication-ready figure showing program impact:

Requirements:

Use the tips dataset
Show how tip percentage varies by day and time (lunch/dinner)
Use faceting or color appropriately
Add proper labels with units
Use a professional theme
Save as both PNG (300 dpi) and PDF

# Your code here

Solution 2

# Prepare data
tips_clean = tips.copy()
tips_clean['tip_percent'] = (tips_clean['tip'] / tips_clean['total_bill']) * 100

# Set theme
sns.set_theme(style="white", context="paper")

# Create plot
plot = (
    so.Plot(tips_clean, x="day", y="tip_percent", color="time")
    .add(so.Dot(alpha=0.3, pointsize=4), so.Jitter(width=0.2))
    .add(so.Dash(width=0.5, linewidth=2), so.Est())
    .add(so.Dot(pointsize=10), so.Agg())
    .scale(color=["#0072B2", "#D55E00"])
    .label(
        title="Tipping Patterns by Day and Meal Time",
        x="Day of Week",
        y="Tip Percentage (%)",
        color="Meal Time"
    )
    .theme({
        "figure.figsize": (8, 5),
        "axes.facecolor": "white",
        "grid.color": "#e5e5e5"
    })
)

# Save
plot.save("tips_analysis.png", dpi=300)
plot.save("tips_analysis.pdf")

plot  # Display

Exercise 3: Complete Figure with Caption

Create a figure for a research report and write an appropriate caption:

Use any dataset (penguins, tips, or your own)
Create a visualization addressing a research question
Apply appropriate theme and styling
Write a complete figure caption (3-5 sentences) that:
- Describes what is shown
- Explains key patterns
- Notes sample size and any important details

Share your figure and caption with a partner for feedback.

Final Tips for Great Visualizations

1. Know Your Audience

Academics: Expect precision, statistical tests, standard formats
Policymakers: Want clear takeaways, simple language, action implications
General public: Need context, plain language, compelling narratives

2. Iterate and Refine

Create a rough version quickly
Get feedback from colleagues
Refine based on feedback
Test with someone unfamiliar with your work

3. Tell a Story

Every figure should have a clear message
Guide the viewer’s eye to key findings
Use titles and captions effectively
Connect visualizations to your narrative

4. Be Honest

Don’t manipulate axes to exaggerate effects
Show uncertainty and variability
Include all relevant data
Report negative findings fairly

Key Points

Set global themes with sns.set_theme(style=, context=)
Fine-tune individual plots with .theme()
Choose themes based on output medium (paper, presentation, poster)
Control figure size with {"figure.figsize": (width, height)}
Save with appropriate format and resolution (.save(filename, dpi=300))
Use dpi=300 for print, dpi=150 for web/presentations
PDF format is best for publications (scalable)
PNG is good for presentations and web
Always include clear titles, labels, and legends
Test your figures with your target audience
Consistency across figures is professional
A polished visualization increases the impact of your research

Congratulations

You’ve completed the Introduction to Data Visualization with Seaborn! You now know how to:

Create visualizations using the grammar of graphics
Map data to visual properties
Use different marks (dots, lines, bars, etc.)
Customize with labels, scales, and colors
Create multi-panel figures with faceting
Layer multiple marks for rich displays
Add statistical summaries
Apply professional themes
Create publication-ready figures

Next steps:

Practice with your own research data
Explore the seaborn documentation
Look at published papers in your field for inspiration
Share your visualizations and get feedback
Keep learning and experimenting!

Themes and Final Polish

The Importance of Visual Polish

Setting Up

Seaborn Themes

Available Themes

Setting Context

Combining Style and Context

Fine-Tuning with .theme()

Common Theme Parameters

Controlling Figure Size

Complete Research Example

Saving Figures

Basic Saving

Resolution Settings

Controlling Output Size

Adding Text and Annotations

Multi-Panel Publication Figure

Best Practices Checklist

✅ Design Principles

Style Guide for Research Visualizations

For Academic Papers

For Presentations

For Policy Briefs

For Posters

Exercises

Final Tips for Great Visualizations

1. Know Your Audience

2. Iterate and Refine

3. Tell a Story

4. Be Honest

The Importance of Visual Polish

Setting Up

Seaborn Themes

Available Themes

Setting Context

Combining Style and Context

Fine-Tuning with .theme()

Common Theme Parameters

Controlling Figure Size

Complete Research Example

Saving Figures

Basic Saving

Resolution Settings

Controlling Output Size

Adding Text and Annotations

Multi-Panel Publication Figure

Best Practices Checklist

✅ Before Sharing Your Visualization

✅ Design Principles

Style Guide for Research Visualizations

For Academic Papers

For Presentations

For Policy Briefs

For Posters

Exercises

Final Tips for Great Visualizations

1. Know Your Audience

2. Iterate and Refine

3. Tell a Story

4. Be Honest