Radar Chart

The radar chart (also called a spider or web chart) draws filled polygons on a circular grid of spokes. Each spoke is a metric, and each polygon is a subject being compared across all those metrics at once. It’s the chart to reach for when you want to see a multi-dimensional profile — not just “which is biggest” but “where does each variant win and where does it lose.”

Use radar charts when your Y axis holds multiple meaningful metrics (throughput, latency, memory, etc.) and you want to compare several subjects across all of them simultaneously.

How vizb builds it

Radar has no cartesian axes. Instead, vizb maps your dimensions to the radial layout:

Dimension	Role
YAxis	Spoke indicators — each unique Y value becomes one spoke radiating from the center
XAxis	Data points — each unique X value becomes a polygon on the web
ZAxis	Series grouping — each Z value becomes a set of polygons; X values become multiple points within each Z polygon (3D only)

Each polygon is filled with a semi-transparent area so overlapping shapes stay visible. Hover any polygon to see a tooltip listing every spoke name alongside its value.

Dimensions

Long-form scores (subject × metric × value). Copy → profiles.csv:

profiles.csvCopy CSV

subject	metric	score	tier
Pond	throughput	92	eager
Pond	latency	70	eager
Pond	memory	80	eager
Pond	stability	88	eager
Chi	throughput	85	eager
Chi	latency	78	eager
Chi	memory	74	eager
Chi	stability	82	eager
Gin	throughput	90	lazy
Gin	latency	65	lazy
Gin	memory	88	lazy
Gin	stability	79	lazy
Echo	throughput	88	lazy
Echo	latency	72	lazy
Echo	memory	81	lazy
Echo	stability	84	lazy

1D (X axis)

One polygon on the radar. The X values become the spokes, and the single data point is plotted as a stat-total shape. Best for a quick profile of one subject.

CLI

vizb radar profiles.csv -g metric -o out.html

HTTP

POST / on vizb serve. The API takes inlineinput, not a file path.

{
  "input": "subject,metric,score,tier\nPond,throughput,92,eager\nPond,latency,70,eager\nPond,memory,80,eager\nPond,stability,88,eager\nChi,throughput,85,eager\nChi,latency,78,eager\nChi,memory,74,eager\nChi,stability,82,eager\nGin,throughput,90,lazy\nGin,latency,65,lazy\nGin,memory,88,lazy\nGin,stability,79,lazy\nEcho,throughput,88,lazy\nEcho,latency,72,lazy\nEcho,memory,81,lazy\nEcho,stability,84,lazy\n",
  "parser": "csv",
  "grouping": {
    "columns": [
      "metric"
    ]
  },
  "charts": {
    "types": [
      "radar"
    ]
  },
  "output": {
    "format": "html"
  }
}

Action

GitHub Action step. Same conversion as the CLI.

- uses: goptics/vizb@v0
  with:
    file: profiles.csv
    group: metric
    charts: radar
    output-html: out.html

2D (X + Y axes)

Multiple polygons, one per X value, all sharing the same set of spokes (Y values). The legend lists the X values so you can show or hide individual polygons. This is the core use case: compare how several subjects score across the same set of metrics.

CLI

vizb radar profiles.csv -g subject,metric -p x,y -o out.html

HTTP

POST / on vizb serve. The API takes inlineinput, not a file path.

{
  "input": "subject,metric,score,tier\nPond,throughput,92,eager\nPond,latency,70,eager\nPond,memory,80,eager\nPond,stability,88,eager\nChi,throughput,85,eager\nChi,latency,78,eager\nChi,memory,74,eager\nChi,stability,82,eager\nGin,throughput,90,lazy\nGin,latency,65,lazy\nGin,memory,88,lazy\nGin,stability,79,lazy\nEcho,throughput,88,lazy\nEcho,latency,72,lazy\nEcho,memory,81,lazy\nEcho,stability,84,lazy\n",
  "parser": "csv",
  "grouping": {
    "columns": [
      "subject",
      "metric"
    ],
    "pattern": "x,y"
  },
  "charts": {
    "types": [
      "radar"
    ]
  },
  "output": {
    "format": "html"
  }
}

Action

GitHub Action step. Same conversion as the CLI.

- uses: goptics/vizb@v0
  with:
    file: profiles.csv
    group: subject,metric
    group-pattern: x,y
    charts: radar
    output-html: out.html

3D (X + Y + Z axes)

Multiple polygon sets, one set per Z value. Within each Z series, X values become multiple data points and Y values remain the spokes. The legend lists the Z values. To keep smaller polygons hoverable, vizb draws the largest Z series first so smaller polygons are rendered on top and their vertices stay reachable.

Spoke vertices are rendered as circle symbols, so you can hover individual points on each polygon to inspect a specific metric value.

CLI

vizb radar profiles.csv -g subject,metric,tier -p x,y,z -o out.html

HTTP

POST / on vizb serve. The API takes inlineinput, not a file path.

{
  "input": "subject,metric,score,tier\nPond,throughput,92,eager\nPond,latency,70,eager\nPond,memory,80,eager\nPond,stability,88,eager\nChi,throughput,85,eager\nChi,latency,78,eager\nChi,memory,74,eager\nChi,stability,82,eager\nGin,throughput,90,lazy\nGin,latency,65,lazy\nGin,memory,88,lazy\nGin,stability,79,lazy\nEcho,throughput,88,lazy\nEcho,latency,72,lazy\nEcho,memory,81,lazy\nEcho,stability,84,lazy\n",
  "parser": "csv",
  "grouping": {
    "columns": [
      "subject",
      "metric",
      "tier"
    ],
    "pattern": "x,y,z"
  },
  "charts": {
    "types": [
      "radar"
    ]
  },
  "output": {
    "format": "html"
  }
}

Action

GitHub Action step. Same conversion as the CLI.

- uses: goptics/vizb@v0
  with:
    file: profiles.csv
    group: subject,metric,tier
    group-pattern: x,y,z
    charts: radar
    output-html: out.html

Caution

Scale (log) and auto-rotate are not available for radar charts. Radar has no cartesian axis to scale logarithmically, and there is no 3D scene to rotate. These controls are hidden from the UI when only a radar chart is active.

Settings

These settings apply to radar charts:

Setting	CLI flag	UI toggle	Notes
Sort	--sort asc|desc	Sort control	Orders polygons / spokes
Labels	--show-labels	Show labels	Displays values at spoke vertices
Swap	--swap	Axis switcher	Swaps which column maps to spokes vs polygons

Override settings for just this chart type without affecting others:

# Sort ascending and show labels
vizb radar profiles.csv -g subject,metric -p x,y --sort asc -l -o out.html

Next Steps

Bar ChartCompare the same data as grouped bars — supports log scale and 3D depth.

Line ChartTrace trends across categories with connected lines — log scale supported.

Pie ChartProportional slices per axis — three pies for a full three-dimensional breakdown.

SettingsAll runtime controls — sort, labels, swap, and more.