Radar Charts in Python

A radar chart (also called a spider chart, spider plot, or radar plot) compares several measurements across the same categorical dimensions. XY places the categories at evenly spaced angles, draws one spoke for each category, and closes every series across the circular seam.

Use radar charts for compact profile comparisons such as product capabilities, model scores, survey dimensions, and operational health. When the angular position is itself a measured quantity rather than a category, start with the polar chart overview instead.

Jump to filled areas or outlines, the data contract, or scale configuration.

Pass the category labels first, followed by one area or line mark per series. Each mark supplies exactly one value per category:

The helper assigns the category labels to the theta axis and closes the first and last category with a straight chord. You do not need to repeat the first value at the end of a series.

Use xy.area(values) for a filled profile or xy.line(values) for an outline. When an existing composition already uses area marks, set fill=False on the chart to turn every area into an outline without rewriting the series:

When fill=False rebuilds an area as a line, the outline inherits line_color, line_width, line_opacity, curve, and dash settings. If line_color is omitted, it falls back to the area's color.

Filled profiles work best with some transparency so overlapping series remain readable. Give each series a name and add xy.legend() when the chart contains more than one profile.

Radar charts have a deliberately narrow input contract:

• Supply at least three categories.
• Every series must contain exactly one value per category.
• Use only area and line marks.
• Put values directly on the mark. Column-name strings are not resolved because the category list supplies the angular positions.

XY raises ValueError for a category/value mismatch instead of silently dropping a dimension or drawing a malformed polygon.

Add xy.r_axis(domain=(minimum, maximum)) when several charts need a common comparison scale. The default radial range begins at zero and extends to the largest value. Use tick_values= for exact rings and label= when the score has a named unit.

The category list owns the theta labels. An authored xy.theta_axis() can customize their style; grid_shape="linear" joins the spokes into polygonal rings, as in the live example. General polar_chart() compositions also accept category strings directly. radar_chart() additionally validates that each series matches the shared category count and closes profiles for you.

Overlay one mark per candidate, pin the radial domain and its rings with xy.r_axis(), and rotate the first category to the top with xy.theta_axis() so repeated benchmarks stay directly comparable:

Radar charts use the shared polar interaction model: hover, radial-only zoom, and reset are available; rotation, box zoom, selection, brushing, and crosshairs are disabled. Browser, SVG, PDF, and native raster exports share the same radar geometry.

See Polar chart interaction and limits for the complete coordinate-system contract.

• Polar overview — measured theta/r data, axes, pyplot compatibility, and shared limits.
• Radial bar charts — magnitudes encoded as annular sectors.
• Wind rose charts — raw compass observations binned by direction and speed.

A radar chart assigns evenly spaced angles to category labels and closes the profile automatically. A polar line chart expects explicit numeric angles and radii.

Call xy.radar_chart(categories, ...) with one xy.area() or xy.line() mark per series and render it:

The category list drives the spokes, so a spider plot needs no manual angle math and no repeated closing value.

Use xy.line(values) for each series, or pass fill=False to xy.radar_chart() when the children are area marks.

Each series in a radar graph must have the same number of values as the category list. Add or remove values so every category has one measurement.