vega/voyager

Voyager 2

Voyager 2 is a data exploration tool that blends manual and automated chart specification. Voyager 2 combines PoleStar, a traditional chart specification tool inspired by Tableau and Polaris (research project that led to the birth of Tableau), with two partial chart specification interfaces: (1) wildcards let users specify multiple charts in parallel,(2) related views suggest visualizations relevant to the currently specified chart. With Voyager 2, we aim to help analysts engage in both breadth-oriented exploration and depth-oriented question answering.

For a quick overview of Voyager, see our preview video, or a 4-minute demo in our Vega-Lite talk at OpenVisConf, or watch our research talk at CHI 2017. For more information about our design, please read our CHI paper and other related papers (1, 2, 3).

Voyager 2 can be used from JupyterLab via the JupyterLab extension for Voyager. The DataVoyager.jl package integrates Voyager 2 into the Julia programming language.

WARNING:

This repository now hosts an alpha version of the migration of Voyager 2 to a React/Redux application. Older versions of Voyager built in AngularJS at the following URL.

• The Voyager 2 visualization tool, which blends manual and automated chart specification – demo at http://vega.github.io/voyager2 and source code at https://github.com/vega/voyager2
• The Voyager 1 visualization browser -- demo at http://uwdata.github.io/voyager and source code in the vy1 branch of this repository.

Documentation

You can find Voyager documentation on our GitBook.

This documentation is divided into several sections:

• Load Data
• Visualizing Data
  • Univariate Summaries
  • Specify Visual Encodings
  • Filter
  • Related Views
  • Wildcards
  • Customizing Visualizations
• Bookmark Gallery
• Using Voyager in JupyterLab

Basic Setup

For basic setup for local development or installation, we use yarn for package management. Installing dependencies can be done with:

yarn

Once the installation is complete, use yarn test to run the included tests.

To build a deployable version of the code, run yarn build.

Please see our contributing documentation for more info about setup and coding conventions if you are interested in contributing to this project.

Build Outputs

There are 3 artifacts build using yarn build:

• Stand alone version of voyager in dist/. This distribution can be hosted on a web server to deploy Voyager.
• Compiled Javscript and .d.js declaration files for a subset of the Voyager source code in build/src/. These declarations and sources can be included in other packages that use Voyager as a dependency. See voyager-server for an example.
• Embeddable Voyager build in build/. See below for more details on embedding Voyager in other applications.

Embed Voyager (datavoyager library)

Voyager can be embedded in another web application. The following sections document how to use it.

Installation

Using npm or yarn? Add the following to your package.json then run npm install datavoyager or yarn add datavoyager.

If you want to use the latest development version, you may want to clone and link Voyager.

Example Use

Instantiation

const libVoyager = require('voyager');

const container = document.getElementById("voyager-embed");
const config = undefined;
const data = undefined;
const voyagerInstance = libVoyager.CreateVoyager(container, config, data)

Initializing with data

const data: any = {
  "values": [
    {"fieldA": "A", "fieldB": 28}, {"fieldA": "B", "fieldB": 55}, {"fieldA": "C", "fieldB": 43},
    {"fieldA": "D", "fieldB": 91}, {"fieldA": "E", "fieldB": 81}, {"fieldA": "F", "fieldB": 53},
    {"fieldA": "G", "fieldB": 19}, {"fieldA": "H", "fieldB": 87}, {"fieldA": "I", "fieldB": 52}
  ]
};

const voyagerInstance = libVoyager.CreateVoyager(container, undefined, data)

Updating Data

const voyagerInstance = libVoyager.CreateVoyager(container, undefined, undefined)

const data: any = {
  "values": [
    {"fieldA": "A", "fieldB": 28}, {"fieldA": "B", "fieldB": 55}, {"fieldA": "C", "fieldB": 43},
    {"fieldA": "D", "fieldB": 91}, {"fieldA": "E", "fieldB": 81}, {"fieldA": "F", "fieldB": 53},
    {"fieldA": "G", "fieldB": 19}, {"fieldA": "H", "fieldB": 87}, {"fieldA": "I", "fieldB": 52}
  ]
};

voyagerInstance.updateData(data);

CSS

You currently also need to include the CSS. Note that this has not yet been optimized for embedding (it will take over the whole screen)

<link rel="stylesheet" type="text/css" href="./node_modules/voyager/lib/style.css">

API

The voyager module exposes 1 function.

CreateVoyager(container, config, data)

/**
 * Create an instance of the voyager application and return it.
 *
 * @param {Container} container css selector or HTMLElement that will be the parent
 *                              element of the application
 * @param {Object|undefined}    config    Optional: configuration options
 * @param {Array|undefined}     data      Optional: data object. Can be a string or an array of objects.
 */

Please see src/lib-voyager.tsx to see the exposed public methods.

For information regarding the config parameter, please see src/models/config.ts

The data parameter must follow the inline data format as seen in the vega lite documentation

Voyager-server Mode

Computationally expensive portions of the Voyager process can be configured to run on a server.

To get this running in a local development environment, first clone and install the dependencies of the voyager-server project.

In voyager-server directory, yarn start will start the server running on port 3000.

With voyager-server now running, we can start voyager in server mode by running:

yarn start:server

This will run Voyager in "server-mode" sending requests to voyager-server, which it expects, by default, to be at http://localhost:3000.

The server url is controlled by the SERVER environment variable.

See voyager-server for more information on what portions of the functionality the server handles.