DataViz.Layout.Hierarchy.Cluster

type ClusterConfig = { minSeparation :: Number, size :: { height :: Number, width :: Number } }

Configuration for cluster layout

defaultClusterConfig :: ClusterConfig

Default configuration

cluster :: forall r. ClusterConfig -> Tree { height :: Int, x :: Number, y :: Number | r } -> Tree { height :: Int, x :: Number, y :: Number | r }

Cluster layout in 4 steps:

1. addHeight: Bottom-up pass computing height (distance from deepest leaf)
2. sortByHeight: Sort children by height to minimize crossovers (D3 pattern)
3. render: Bottom-up pass assigning sequential x to leaves, mean x to parents
4. addY: Add y coordinates based on height (leaves at 0)
5. scale: Scale abstract coordinates to pixel coordinates

Key difference from Tree:

• All leaves get sequential x positions (not contour-based)
• y is based on height (distance from leaves) not depth
• All leaves appear at y = 0 (dendrogram)
• Children sorted by height (descending) to minimize crossovers

Input must have x, y, height fields (initial values don't matter, they'll be overwritten)