Skip to content

graphiquejs/graphique

Repository files navigation

An interactive visualization system for React based on the Grammar of Graphics


Graphique

Graphique allows you to concisely create flexible and reusable interactive visualizations by:

  • mapping variables in data to visual components or aesthetics specified by scales (how to draw)
  • composing or layering relatively simple graphical objects or geometries (what to draw)
  • using reasonable defaults that can be configured for flexibility

Examples/Docs

Examples and docs: https://graphique.dev

Usage

Creating a simple, interactive line chart

Install only the geometries you need.

npm install @graphique/graphique @graphique/geom-line @graphique/datasets

or

yarn add @graphique/graphique @graphique/geom-line @graphique/datasets
import React from 'react'
import ReactDOM from 'react-dom'
import { GG } from '@graphique/graphique'
import { GeomLine } from '@graphique/geom-line'
import { stocks } from '@graphique/datasets'

// default basic line chart
const LineChart = () => {
  return (
    <GG
      data={stocks.filter((d) => d.symbol === 'AAPL')}
      aes={{
        x: (d) => new Date(d.date),
        y: (d) => d.marketCap,
      }}
    >
      <GeomLine />
    </GG>
  )
}

ReactDOM.render(<LineChart />, document.getElementById('root'))

Responsive

To keep your Graphique visualization the width of its parent container, you only need to specify isContainerWidth and you'll get an appropriately-scaled and responsive version.

<GG
  data={stocks.filter((d) => d.symbol === 'AAPL')}
  aes={{
    x: (d) => new Date(d.date),
    y: (d) => d.marketCap,
  }}
  isContainerWidth
>
  <GeomLine />
</GG>

Design philosophy

Relative to existing data visualization tools and libraries for React applications, Graphique exists "above" low-level APIs and "below" high-level/out-of-the-box/"named chart" charting libraries.

As much as possible, the routine parts of visualizing data should be taken care of for you. This means (by default) you shouldn't need to manually create your own scales, axes, coordinate systems, tooltips, interactive legends, animated transitions, or responsive logic. But when it comes time, you can customize/override nearly everything to suit your specific needs.

With sensible defaults, Graphique aims to be just opinionated enough to make the process of creating highly customizable visualizations as streamlined as possible. Graphique is directly inspired by ggplot2, and under the hood it's built on d3.

A layered approach

// empty / no geoms
// defaults provide:
// x/y scales, coordinate system,
// grid, axes, ticks, dimensions, theme
<GG
  data={stocks}
  aes={{
    x: (d) => new Date(d.date),
    y: (d) => d.marketCap,
  }}
/>

Aesthetics/Scales

How to draw. The specifications for how data characteristics relate to visual characteristics (functional mappings/transformations applied to data).

Aesthetic (Scale) Specifies Notes
x (ScaleX) x positions of geometry objects, axis creation
y (ScaleY) y positions of geometry objects, axis creation
fill (ScaleFill) fill color of geometry objects, groups if categorical
stroke (ScaleStroke) stroke color of geometry objects, groups if categorical
strokeDasharray (ScaleDashArray) dash pattern of geometry objects, groups categorical only
size (ScaleSize) size of geometry objects continuous only, GeomPoint only
group (n/a) explicit groups for geometry objects
label (n/a) main label for geometry object tooltips
key (n/a) uniqueness/identity of geometry objects

You can add one or more Scales directly to customize things like its type (d3 scale), domain, values, formatting, etc.

// a log-transformed y scale with
// custom domain and formatted as USD
<GG
  data={data}
  aes={{
    x: (d) => d.xVal,
    y: (d) => d.yVal,
  }}
>
  <ScaleY
    type={d3.scaleLog}
    domain={[0, 100000]}
    format={({ value }) => `$${value.toLocaleString()}`}
  />
</GG>

Geometries

What to draw. The shapes that give a visualization something to visualize.

  • GeomLine: line charts and other kinds of lines
  • GeomPoint: scatterplots, dotplots, and bubble charts
  • GeomTile: rectangular charts like heatmaps
  • GeomBar: bar charts (horizontal bars)
  • GeomCol: column charts (vertical bars)
  • GeomHist: histograms
  • GeomArea: area + stacked area charts and ribbon-like things
  • GeomVLine: vertical lines / markers
  • GeomHLine: horizontal lines / markers
  • TODO: GeomLabel, GeomSmooth, GeomSegment, GeomDensity, and more on the way!

Labels

Give the main parts flexible, human-readable labels.

  • header
  • x
  • y

Legend

Each Geom includes a <Legend> that can be added to your Graphique visualization. Continuous aes<->data return values get continuous legends. Categorical aes<->data return values get categorical legends.

Tooltip

Tooltips are provided for each Geom for free. They're configurable and you can roll your own based on the relevant contextual information (x value, y value, etc) with <Tooltip>. If you'd prefer the Geom to not use a tooltip, you can turn it off by passing showTooltip={false} to the Geom.

Theme

Customize the look and feel of your Graphique visualizations.

// a custom theme
<Theme
  grid={{ stroke: 'none' }}
  axis={{ tickStroke: '#ddd' }}
  axisX={{ showAxisLine: true }}
  font={{ family: 'Inter, system-ui, sans-serif' }}
/>

Roadmap

  • more Geoms
  • linked graphics inside <GGgroup>

Dev

Testing

npm test

Run only some tests

For example, you can also choose to run only some tests with things like:

# run only GeomLine tests
npm test GeomLine

# run only Tooltip tests (multiple Geoms)
npm test Tooltip

Demo local packages

For a development sandbox / place to try out new package changes locally, there's a Vite app in ./demo.

You can start the demo app with:

npm run demo

Publishing and releasing

Use npm run release to publish and release new version(s) of package(s).