Australian Federal Electorate Hex Cartogram

Equal-area hex maps for Australian federal House of Representatives electorates, covering the 2019, 2022, and 2025 elections.

Each electorate is represented as a single hexagon on a unified grid. The layout approximates Australia's geography while giving every electorate exactly the same visual weight — unlike geographic maps where outback electorates dwarf inner-city ones.

Based on the ABC News federal election hex map, which pioneered the equal-area hex cartogram format for Australian elections.

Install

npm install auspol-hex-cartogram

Quick start (React)

import { FED_2025 } from "auspol-hex-cartogram";
import { Cartogram } from "auspol-hex-cartogram/react";

function App() {
  return (
    <Cartogram
      election={FED_2025}
      fill={(e) => e.state === "VIC" ? "#1D4ED8" : "#6B7280"}
      onElectorateClick={(e) => console.log(e.name)}
    />
  );
}

Quick start (vanilla JS)

import { FED_2022, HEX_SIZE, hexPoints, computeStateBorders } from "auspol-hex-cartogram";

const { electorates } = FED_2022; // 151 electorates

for (const e of electorates) {
  // e.name = "Kooyong", e.state = "VIC", e.col = 7, e.row = 14
  // e.px, e.py = pre-computed pixel coordinates for SVG rendering
  const points = hexPoints(e.px, e.py, HEX_SIZE);
  // → SVG polygon points string
}

// SVG path string for state boundary lines
const borderPath = computeStateBorders(electorates, HEX_SIZE);

Or load the raw JSON directly:

import data from "auspol-hex-cartogram/data/elections.json";
const grid = data["2025-FED"].grid; // [[col, row, state, name], ...]

Data format

All election data lives in data/elections.json. Each entry is a [col, row, state, name] tuple:

[6, 13, "VIC", "Kooyong"]

The grid uses pointy-top hexagons with odd-row offset:

px = √3 × size × (col + 0.5 × (row & 1))
py = 1.5 × size × row

where size = 14 pixels. Grid bounds: col 0–13, row 0–19.

API reference

Types

| Type | Description | |------|-------------| | GridEntry | Raw tuple: [col, row, state, name] | | Electorate | Resolved object: { code, name, state, seatId, col, row, px, py } | | ElectionMap | Dataset: { electionId, label, seatCount, grid, electorates } |

Functions

| Function | Description | |----------|-------------| | cellToPixel(col, row) | Grid coords → { x, y } pixel centre | | hexPoints(cx, cy, size) | Pixel centre → SVG polygon points string | | computeStateBorders(hexes, size) | Array of electorates → SVG d path for state borders | | resolveGrid(grid) | Raw GridEntry[] → Electorate[] with computed pixels | | nameToSeatId(name) | "Kingsford Smith" → "KINGSFORDSMITH" |

Constants

| Constant | Value | Description | |----------|-------|-------------| | HEX_SIZE | 14 | Default hex radius in pixels | | STATE_HEX_COLORS | Record<string, string> | State → CSS colour map |

Election datasets

| Export | Election | Seats | |--------|----------|-------| | FED_2019 | 2019 Federal | 151 | | FED_2022 | 2022 Federal | 151 | | FED_2025 | 2025 Federal | 150 |

React components

Import from auspol-hex-cartogram/react:

<Cartogram> — full SVG map with state borders and hover effects.

| Prop | Type | Default | Description | |------|------|---------|-------------| | election | ElectionMap | FED_2022 | Which election to render | | fill | (e: Electorate) => string | State colours | Colour callback per electorate | | tooltip | (e: Electorate) => string | "Name (STATE)" | Tooltip text callback | | onElectorateClick | (e: Electorate) => void | — | Click handler | | showBorders | boolean | true | Show state border lines | | hexSize | number | 14 | Hex radius in pixels | | style | CSSProperties | — | SVG inline styles | | className | string | — | SVG class name |

<ElectorateCell> — single hex polygon with native SVG tooltip.

Elections

| Election | Seats | Changes from previous | |----------|-------|-----------------------| | 2019 Federal | 151 | Derived backward from 2025 via 2022 | | 2022 Federal | 151 | +Hawke (VIC), −Stirling (WA) | | 2025 Federal | 150 | Canonical baseline (matches ABC News) |

Silhouette changes

2019 → 2022: VIC gains cell [5,12] for Hawke (Wannon and Corangamite shift to fill it). WA loses cell [1,7] (Stirling abolished). 148 surviving electorates stay put; 2 forced moves.

2022 → 2025: NSW loses [13,10] (North Sydney abolished). VIC loses [7,16] (Higgins abolished). WA gains [2,11] for Bullwinkel (Curtin, Cowan, and Perth shift to accommodate). 146 surviving electorates stay put; 3 forced moves.

How the coordinates were derived

2025: canonical baseline (from ABC News)

The 2025 positions are taken directly from the ABC News 2025 federal election hex map (abcnews/elections-federal2025-lower-house on GitHub). The ABC's per-state local coordinates are combined with their COUNTRY layout state offsets to produce a unified integer grid.

This is the canonical reference layout. All other elections are derived from it.

2022 and 2019: stability-first derivation (backwards)

When electorates are added or removed between elections, we use a stability-first approach, working backwards from the 2025 baseline:

1. Start from the 2025 baseline positions.
2. Abolished electorates (seats that existed in the earlier election but not 2025): add them back on a new cell adjacent to their state's silhouette.
3. New electorates (seats that exist in 2025 but not the earlier election): remove them and their cell.
4. Surviving electorates keep their position unless a forced move is needed to avoid leaving a hole in the silhouette. Forced moves cascade from the hole to the nearest edge cell.

This produces minimal visual disruption between elections — critical for comparing results across years.

| Transition | Stayed | Moved | New | Removed | |------------|--------|-------|-----|---------| | 2019 → 2022 | 148 | 2 | 1 (Hawke) | 1 (Stirling) | | 2022 → 2025 | 146 | 3 | 1 (Bullwinkel) | 2 (Higgins, North Sydney) |

Why not the Hungarian algorithm?

We investigated using the Hungarian algorithm to optimally assign electorates to hex cells based on geographic centroids. The results were instructive:

Geographic accuracy: Hungarian wins slightly

| Metric | Hand-crafted | Hungarian | Difference | |--------|-------------|-----------|------------| | RMS positional error | 69.2 px | 61.5 px | +12.5% | | Mean error | 58.1 px | 52.3 px | +11.1% | | Max error | 201.5 px | 143.4 px | +40.6% |

The hand-crafted layout is ~12% worse on average — roughly 0.3 hex-widths per electorate. For a cartogram that inherently distorts geography (an electorate covering half of Western Australia is the same size as one covering 3 km² of inner Melbourne), this is a reasonable trade-off.

Stability: Hungarian fails catastrophically

When run independently per election, the Hungarian algorithm produced 113 unnecessary moves between 2019 and 2022 — reshuffling nearly every electorate even though only 2 seats changed.

| Transition | Stability-first | Hungarian | |------------|----------------|-----------| | 2019 → 2022 moves | 0 | 113 | | Unnecessary moves | 0 | 112 |

This happens because the algorithm optimises each election in isolation. A tiny shift in centroid positions cascades into swaps across entire states.

Our decision

Stability over optimality. The 12% geographic error is acceptable. The 113-electorate disruption is not. Users comparing election results across years need electorates in predictable positions.

The Hungarian algorithm remains useful for validation (checking the hand-crafted layout isn't wildly wrong) and for initial placement of new electorates. But for cross-election consistency, stability-first wins.

Relationship to the ABC

Our 2025 baseline is taken directly from the ABC's hex map (source). The coordinates are identical for the 2025 election.

The ABC also published a hex map for the 2022 election, but our 2022 layout differs because it is derived backwards from 2025 using stability-first (all surviving electorates keep their 2025 positions). The ABC has not published a hex map for 2019.

Adding a new election

After a federal redistribution:

1. Identify added, removed, and renamed electorates.
2. For each state that gained/lost seats, decide which cells to add/remove from the silhouette.
3. Apply the stability-first rule: surviving electorates keep their positions.
4. For forced moves, prefer adjacent cells. Use Hungarian within the affected state if there are many.
5. Add the new election to data/elections.json.
6. Run tests: npm test

Examples

• examples/vanilla.html — simple interactive cartogram with election switcher
• examples/analysis.html — transition analysis showing movements, geographic error, and state seat counts

License

MIT