position-beeswarm
Density-shaped per-row offset on x.
For every bucket of rows sharing an x value, each point is offset sideways by width × (2·v − 1) × d, where v walks the van der Corput base-2 sequence in y order and d is the point’s Gaussian kernel density estimate normalised to the bucket’s peak. Dense regions spread wide, sparse tails stay close to the spine, and the result is a violin-shaped swarm.
A discrete (categorical) x is swarmed directly on its level positions, keeping the axis labels. as-factor is only needed to force a numeric column onto a discrete axis.
Usage
position-beeswarm(
width: 0.4,
adjust: 1,
)Parameters
| Parameter | Default | Description |
|---|---|---|
width |
0.4 |
Maximum absolute offset applied to the x position, in x data units. |
adjust |
1 |
Bandwidth multiplier for the density estimate: adjust: 0.5 halves the smoothing. |
Returns
Position dictionary with name: "beeswarm", consumed by plot.
Examples
Swarm overplotted points per group; a tighter width keeps the cloud close to its spine.
#let d = ()
#for grp in ("a", "b") {
for i in range(0, 50) {
d.push((grp: grp, y: calc.sin(i * 0.7) + calc.sin(i * 1.9) + (if grp == "b" { 4 } else { 0 })))
}
}
#plot(
data: d,
mapping: aes(x: as-factor("grp"), y: "y"),
layers: (geom-point(size: 2pt, position: position-beeswarm(width: 0.25)),),
width: 10cm,
height: 6cm,
)
