Logo API#

Scikit-plots includes a small brand helper that generates the official logo using pure Matplotlib primitives. The default output is deterministic and requires no image assets.

This makes the logo easy to:

ship inside the library,
regenerate at any size,
theme for dark/light docs,
export as SVG/PNG without external dependencies.

Note

The public interface is exposed as _logo. The internal implementation may live in _logo.

If you are developing inside the repository you may see examples using sp._logo; user code should prefer sp._logo when available.

Quickstart#

import scikitplot as sp

# Show primary icon
sp._logo.show()

# Save a single icon
sp._logo.save("scikit-plots.svg")

# Save multiple variants using a template
sp._logo.save(
    "assets/scikit-plots-{variant}.svg",
    variants=sp._logo.list_variants()
)

Variants#

Scikit-plots provides several icon variants tuned for different contexts:

primary: full brand mark for docs, README, websites.
small: simplified icon for tiny sizes.
metrics: highlights classic ML evaluation concepts.
knn: subtle nod to Annoy/KNN features.

sp._logo.list_variants()
# ('primary', 'small', 'metrics', 'knn')

# Save all variants with auto suffix
sp._logo.save("assets/scikit-plots.svg", variants=sp._logo.list_variants())

Themes & Monochrome#

Use themes for dark backgrounds and monochrome for badges or watermarks.

# Dark theme for docs landing pages
sp._logo.save("assets/scikit-plots-dark.svg", theme="dark")

# Monochrome for badges
sp._logo.save("assets/scikit-plots-mono.svg", mono=True)

Decorative Dots#

The primary mark supports a small dot field. For brand-stable output, keep dots="fixed" (the default).

# Brand-stable default
sp._logo.save("assets/scikit-plots-primary.svg", dots="fixed")

# Turn off dots for extra minimal output
sp._logo.save("assets/scikit-plots-clean.svg", dots="none")

# Experimental random layout (not recommended for official assets)
sp._logo.save("assets/scikit-plots-random.svg", dots="random", seed=42)

Presets#

Presets provide convenient defaults for common contexts such as favicons, avatars, and documentation hero images.

sp._logo.list_size_presets()
# ('favicon', 'avatar', 'docs-hero')

# A crisp small icon for browser/app use
sp._logo.save("assets/favicon.png", preset="favicon")

# A balanced default for social/GitHub avatars
sp._logo.save("assets/avatar.svg", preset="avatar")

# A large icon for docs landing pages
sp._logo.save("assets/hero.svg", preset="docs-hero")

Wordmark#

The wordmark combines the icon with a text lockup. This is useful for website headers, slides, and banners.

# Show wordmark
sp._logo.wordmark.show()

# Save wordmark lockup
sp._logo.wordmark.save("assets/scikit-plots-lockup.svg")

Saving Behavior#

The save API follows Matplotlib-style conventions:

Format is inferred from filename suffix.
Override inference with ext= or format=.
Extra keyword arguments are forwarded to Figure.savefig.

# Infer format from suffix
sp._logo.save("assets/scikit-plots.png", dpi=300)

# Force SVG even without suffix
sp._logo.save("assets/scikit-plots", format="svg")

# Save multiple variants using a template
sp._logo.save(
    "assets/scikit-plots-{variant}.png",
    variants=["primary", "small"],
    dpi=300
)

Command Line#

A lightweight CLI is available for batch generation.

# Export all icon variants
python -m scikitplot._logo --all --out assets

# Generate a favicon preset
python -m scikitplot._logo --all --preset favicon --out assets --format png
python -m scikitplot._logo --preset favicon --out assets --format png

# Generate the wordmark lockup
python -m scikitplot._logo --wordmark --out assets --format svg
python -m scikitplot._logo --wordmark --out assets --format png

Developing & Testing Logos#

When extending or adjusting the brand system:

Prefer testing artist-level determinism rather than raw SVG bytes.
Keep dots="fixed" for official outputs.
Add new variants by composing existing primitives (bars, wedges, lines).

If you need a backend-stability check for SVG serialization, compare normalized SVG output rather than byte-for-byte equality.