HMPlot

Treemap

A hierarchy drawn as nested rectangles, each leaf's area set by its value.

Rectangles, colored by depth. x from 0 to 1, y from 0 to 1. 8 rectangles.

Static preview; loading the interactive plot

import hmplot as hm

data = {"source": ["root", "root", "a", "a", "a", "b", "b"],
        "target": ["a", "b", "a1", "a2", "a3", "b1", "b2"]}
with hm.Plot(data=data, height=300) as plot:
    with hm.GraphData(source="source", target="target"):
        with hm.Treemap(root="root", padding=0.02):
            with hm.Nodes():
                hm.Rect(x0="x0", x1="x1", y0="y0", y1="y1", fill="depth")
plot

Options

Keywords are snake_case.

padding_innernumber
The gap between sibling rectangles. Default: padding. In data units.
padding_outernumber
The gap between a parent's edges and its children. Default: padding. In data units.
tile"squarify" | "resquarify" | "binary" | "slice" | "dice" | "sliceDice"
How each rectangle is split among its children, by d3's tiling methods. Default: "squarify".
rootstring | number | boolean
The ID of the node whose subtree the layout divides: the hierarchy's root, or any node below it (a drill-down), or an input a Select or a click sets. Nodes outside its subtree stay rows of the nodes table, hidden; depth counts from it (0).
levelsnumber
How many levels below root to show (1: its children only); root itself is then left out, as its area is the whole layout. Default: every level.
valuecolumn | expression
Each leaf's nonnegative weight, which sets its rectangle's area: a nodes-table column or an expression over it. A parent holds the sum of its leaves. Default: 1.
widthnumber
The width of the rectangle the layout divides. Default: 1 data units.
heightnumber
The height of the rectangle the layout divides. Default: 1 data units.
paddingnumber
The gap around and between rectangles, where padding_inner and padding_outer do not set their own. Default: 0 data units.

Usage

Treemap uses d3-hierarchy to subdivide one connected, directed hierarchy. Supply the edges (and nodes, or let it infer them) and endpoint mapping through GraphData; set root on Treemap. value evaluates on node columns; leaf values determine area and default to 1. width and height default to 1. padding, paddingInner, and paddingOuter control gaps. tile selects squarify (default), resquarify, binary, slice, dice, or sliceDice.

Inside <Nodes>, x0, x1, y0, and y1 give rectangles; x, y, and position give centers. parent, depth, degree, and node attributes are also available. <Edges> retains source rows and exposes center-to-center start and end.

Drilling down

root may be any node, not only the hierarchy’s: the layout divides that node’s subtree to fill the whole area, and the other nodes stay rows of the nodes table, hidden, so marks draw only the subtree. depth counts from root (0), and levels={1} shows its children only (root itself is then left out, as its rectangle is the whole area). Each node’s total is its summed value, for labels and lists; parent is its parent’s ID, for an Up control. With const root = input("root", { default: "all" }) and root={root}, a click on a rectangle can set the input: select={root.set("id")} on the Rect writes the clicked node’s ID, a click on nothing restores the default, a Button with set={root.reset()} goes back to the top, and a Select named root shows and edits the same value. A node table may not have a column named total.

<script>
  const root = input("root", { default: "all" });
</script>

<InputContext>
  <Select name="root" options={["all", ...divisions]} />
  <Button label="Top" set={root.reset()} />
  <Plot>
    <GraphData {nodes} {edges} source="parent" target="id">
      <Treemap root={root} value="revenue" levels={1} padding={0.004}>
        <Nodes bind:this={shown}>
          <Rect x0="x0" x1="x1" y0="y0" y1="y1" fill="division" select={root.set("id")} />
          <Text x="x" y="y" text="id" />
        </Nodes>
      </Treemap>
    </GraphData>
  </Plot>
</InputContext>
root = hm.input("root", default="all")
with hm.Scene() as scene:
    hm.Select(name="root", options=["all", *divisions])
    hm.Button(label="Top", set=root.reset())
    with hm.Plot():
        with hm.GraphData(nodes=nodes, edges=edges, source="parent", target="id"):
            with hm.Treemap(root=root, value="revenue", levels=1):
                with hm.Nodes():
                    hm.Rect(x0="x0", x1="x1", y0="y0", y1="y1", fill="division", select=root.set("id"))

await shown.read({ columns: ["id", "total"] }) gives the shown nodes for a breadcrumb or a list, or put a TableView inside the Nodes, after a Sort for a top ten.

Related examples

Treemap