Python Dataviz

Python Data Visualization

Python data visualization with matplotlib and seaborn for creating publication-quality figures, statistical graphics, and exploratory visualizations.

When to use each library

Matplotlib is the foundational plotting library. Use it for:

• Fine-grained control over every plot element
• Custom layouts with GridSpec or subplot_mosaic
• 3D visualizations
• Animations
• Embedding plots in GUI applications
• When you need low-level customization

Seaborn builds on matplotlib for statistical visualization. Use it for:

• Statistical plots with automatic aggregation and confidence intervals
• Dataset-oriented plotting from DataFrames
• Faceted multi-panel figures (small multiples)
• Distribution visualization (KDE, histograms, violin plots)
• Correlation matrices and clustered heatmaps
• Publication-ready aesthetics with minimal code

Combined approach: Use seaborn for the main visualization, then customize with matplotlib.

Core concepts

Matplotlib hierarchy

1. Figure - Top-level container for all plot elements
2. Axes - Actual plotting area (one Figure can have multiple Axes)
3. Artist - Everything visible (lines, text, ticks, patches)
4. Axis - The x/y number lines with ticks and labels

Two matplotlib interfaces

Object-oriented interface (recommended):

import matplotlib.pyplot as plt

fig, ax = plt.subplots(figsize=(10, 6))
ax.plot(x, y, linewidth=2, label='data')
ax.set_xlabel('X Label')
ax.set_ylabel('Y Label')
ax.legend()
plt.savefig('figure.png', dpi=300, bbox_inches='tight')

pyplot interface (quick exploration only):

plt.plot(x, y)
plt.xlabel('X Label')
plt.show()

Always use the object-oriented interface for production code.

Seaborn function levels

Axes-level functions plot to a single matplotlib Axes:

• Accept ax= parameter for placement
• Return Axes object
• Examples: scatterplot, histplot, boxplot, heatmap

Figure-level functions manage entire figures with faceting:

• Use col, row parameters for small multiples
• Return FacetGrid, JointGrid, or PairGrid objects
• Cannot be placed in existing figures
• Examples: relplot, displot, catplot, lmplot, jointplot, pairplot

import seaborn as sns

# Axes-level: integrates with matplotlib
fig, axes = plt.subplots(1, 2)
sns.scatterplot(data=df, x='x', y='y', ax=axes[0])
sns.histplot(data=df, x='x', ax=axes[1])

# Figure-level: automatic faceting
sns.relplot(data=df, x='x', y='y', col='category', hue='group')

Seaborn semantic mappings

Map data variables to visual properties automatically:

• hue - Color encoding
• size - Point/line size
• style - Marker/line style
• col, row - Facet into subplots

sns.scatterplot(data=df, x='x', y='y',
                hue='category',      # Color by category
                size='importance',   # Size by value
                style='type')        # Different markers

Quick start workflow

1. Import libraries

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

2. Set theme (optional)

sns.set_theme(style='whitegrid', context='paper', font_scale=1.1)

3. Create the plot

# Simple seaborn plot
fig, ax = plt.subplots(figsize=(10, 6))
sns.scatterplot(data=df, x='total_bill', y='tip', hue='day', ax=ax)

# Or figure-level with faceting
g = sns.relplot(data=df, x='x', y='y', col='category', kind='scatter')

4. Customize with matplotlib

ax.set_xlabel('Total Bill ($)', fontsize=12)
ax.set_ylabel('Tip ($)', fontsize=12)
ax.set_title('Restaurant Tips', fontsize=14)
ax.legend(title='Day', bbox_to_anchor=(1.05, 1))

5. Save the figure

plt.savefig('figure.png', dpi=300, bbox_inches='tight')
plt.savefig('figure.pdf')  # Vector format for publications

Plot type selection

For detailed plot type examples, see references/plot-types.md.

Best practices

Interface and layout

1. Use object-oriented interface - Explicit control, easier debugging
2. Use constrained_layout=True - Prevents overlapping elements
3. Set figsize at creation - fig, ax = plt.subplots(figsize=(10, 6))
4. Close figures explicitly - plt.close(fig) to prevent memory leaks

Data preparation

1. Use tidy/long-form data - Each variable a column, each observation a row
2. Use meaningful column names - Seaborn uses them as axis labels
3. Pass DataFrames - Not raw arrays, to preserve semantic information

Color and accessibility

1. Use perceptually uniform colormaps - viridis, plasma, cividis
2. Avoid rainbow colormaps - jet is not perceptually uniform
3. Consider colorblind users - Use viridis, cividis, or colorblind palette
4. Use diverging colormaps for centered data - coolwarm, RdBu for data with meaningful zero

Export

1. Use 300 DPI for publications - dpi=300
2. Use vector formats for print - PDF, SVG
3. Use bbox_inches='tight' - Removes excess whitespace
4. Set explicit figure size - Control dimensions in inches

Statistical plots

1. Understand automatic aggregation - Seaborn computes means and CIs by default
2. Specify error representation - errorbar='sd', errorbar=('ci', 95)
3. Show individual data points - Combine stripplot with boxplot

Common patterns

Multi-panel figure

fig, axes = plt.subplots(2, 2, figsize=(12, 10), constrained_layout=True)
sns.scatterplot(data=df, x='x', y='y', ax=axes[0, 0])
sns.histplot(data=df, x='x', ax=axes[0, 1])
sns.boxplot(data=df, x='cat', y='y', ax=axes[1, 0])
sns.heatmap(corr_matrix, ax=axes[1, 1], cmap='coolwarm', center=0)

Publication figure

sns.set_theme(style='ticks', context='paper', font_scale=1.1)
fig, ax = plt.subplots(figsize=(8, 6))

sns.boxplot(data=df, x='treatment', y='response', ax=ax)
sns.stripplot(data=df, x='treatment', y='response', color='black', alpha=0.3, ax=ax)

ax.set_xlabel('Treatment Condition')
ax.set_ylabel('Response (units)')
sns.despine()

plt.savefig('figure.pdf', dpi=300, bbox_inches='tight')

Faceted exploration

g = sns.relplot(
    data=df, x='x', y='y',
    hue='treatment', style='batch',
    col='timepoint', col_wrap=3,
    kind='line', height=3, aspect=1.5
)
g.set_axis_labels('X Variable', 'Y Variable')
g.set_titles('{col_name}')

Scripts

This skill includes helper scripts:

• scripts/plot_template.py - Template demonstrating various plot types
• scripts/style_configurator.py - Interactive style configuration utility

References

For detailed information, load specific references:

oaps skill context python-dataviz --references <name>