Skip to content

Latest commit

 

History

History
815 lines (587 loc) · 26.4 KB

README.md

File metadata and controls

815 lines (587 loc) · 26.4 KB

LineUp.js: Visual Analysis of Multi-Attribute Rankings

License NPM version CircleCI CircleCI (dev)

LineUp is an interactive technique designed to create, visualize and explore rankings of items based on a set of heterogeneous attributes.

Key Features

  • scalable (~1M rows)
  • heterogenous attribute types (string, numerical, categorical, boolean, date)
  • composite column types (weighted sum, min, max, mean, median, impose, nested, ...)
  • array (multi value) and map column types (strings, stringMap, numbers, numberMap, ...)
  • filtering capabilities
  • hierarchical sorting (sort by more than one sorting criteria)
  • hierarchical grouping (split rows in multiple separate groups)
  • group aggregations (show a whole group as a single group row)
  • numerous visualizations for summaries, cells, and group aggregations
  • side panel for easy filtering and column management
  • React, Angular, Vue.js, Polymer, RShiny, Juypter, ObservableHQ, and Power BI wrapper
  • Demo Application with CSV import and export capabilities
  • API Documentation based on generated TypeDoc documenation

Usage

Installation

npm install --save lineupjs
<link href="https://unpkg.com/lineupjs/build/LineUpJS.css" rel="stylesheet">
<script src="https://unpkg.com/lineupjs/build/LineUpJS.js"></script>

Minimal Usage Example

// generate some data
const arr = [];
const cats = ['c1', 'c2', 'c3'];
for (let i = 0; i < 100; ++i) {
  arr.push({
    a: Math.random() * 10,
    d: 'Row ' + i,
    cat: cats[Math.floor(Math.random() * 3)],
    cat2: cats[Math.floor(Math.random() * 3)]
  })
}
const lineup = LineUpJS.asLineUp(document.body, arr);

CodePen

Minimal Result

Advanced Usage Example

// arr from before
const builder = LineUpJS.builder(arr);

// manually define columns
builder
  .column(LineUpJS.buildStringColumn('d').label('Label').width(100))
  .column(LineUpJS.buildCategoricalColumn('cat', cats).color('green'))
  .column(LineUpJS.buildCategoricalColumn('cat2', cats).color('blue'))
  .column(LineUpJS.buildNumberColumn('a', [0, 10]).color('blue'));

// and two rankings
const ranking = LineUpJS.buildRanking()
  .supportTypes()
  .allColumns() // add all columns
  .impose('a+cat', 'a', 'cat2'); // create composite column
  .groupBy('cat')
  .sortBy('a', 'desc')
  

builder
  .defaultRanking()
  .ranking(ranking);

const lineup = builder.build(document.body);

CodePen

Advanced Result

Supported Browsers

  • Chrome 64+ (best performance)
  • Firefox 57+
  • Edge 16+

Demo Application

A demo application is located at lineup_app. It support CSV Import, CSV Export, JSON Export, CodePen Export, nad local data management.

The application is deployed at https://lineup.js.org/app

Screenshot

API Documentation

LineUp is implemented in clean TypeScript in an object oriented manner. A fully generated API documentation based on TypeDoc is available at https://lineup.js.org/master/docs

LineUp can be build manually or using via the builder design pattern (see Advanced Usage Example). The builder design pattern in the more common way.

LineUp Builder

The simplest methods to create a new instance are:

  • asLineUp returning a ready to use LineUp instance
    asLineUp(node: HTMLElement, data: any[], ...columns: string[]): LineUp
  • asTaggle returning a ready to use Taggle instance
    asTaggle(node: HTMLElement, data: any[], ...columns: string[]): Taggle
  • builder returning a new DataBuilder
    builder(arr: any[]): DataBuilder`

The DataBuilder allows on the one hand to specify the individual columns more specificly and the creation of custom rankings.

Builder factory functions for creating column descriptions include:

In order to build custom rankings within the DataBuilder the buildRanking returning a new RankingBuilder is used.

buildRanking(): RankingBuilder

LineUp classes and manual creation

The relevant classes for creating a LineUp instance manually are LineUp, Taggle, and LocalDataProvider. A LocalDataProvider is an sub class of ADataProvider implementing the data model management based on a local JavaScript array. LineUp and Taggle are the visual interfaces to the LocalDataProvider.

The classes can be instantiated either using the factory pattern or via their regular class constructors:

createLineUp(container: HTMLElement, data: ADataProvider, config?: Partial<ILineUpOptions>): LineUp

createTaggle(container: HTMLElement, data: ADataProvider, config?: Partial<ITaggleOptions>): Taggle

createLocalDataProvider(data: any[], columns: IColumnDesc[], options?: Partial<ILocalDataProviderOptions>): LocalDataProvider
new LineUp(node: HTMLElement, data: DataProvider, options?: Partial<ILineUpOptions>): LineUp
new Taggle(node: HTMLElement, data: DataProvider, options?: Partial<ITaggleOptions>): Taggle
new LocalDataProvider(data: any[], columns?: IColumnDesc[], options?: Partial<ILocalDataProviderOptions & IDataProviderOptions>): LocalDataProvider

Both LineUp and Taggle are sub classes of ALineUp. The most important functions of this class include:

React Support (LineUp.jsx)

A React wrapper is located at lineupjsx.

Installation

npm install --save lineupjsx
<link href="https://unpkg.com/lineupjsx/build/LineUpJSx.css" rel="stylesheet">
<script src="https://unpkg.com/lineupjsx/build/LineUpJSx.js"></script>

Minimal Usage Example

// generate some data
const arr = [];
const cats = ['c1', 'c2', 'c3'];
for (let i = 0; i < 100; ++i) {
  arr.push({
    a: Math.random() * 10,
    d: 'Row ' + i,
    cat: cats[Math.floor(Math.random() * 3)],
    cat2: cats[Math.floor(Math.random() * 3)]
  })
}
<LineUp data={arr}/>

CodePen

Result is same as the builder minimal example

Advanced Usage Example

// arr from before
<LineUp data={arr} defaultRanking>
  <LineUpStringColumnDesc column="d" label="Label" width={100} />
  <LineUpCategoricalColumnDesc column="cat" categories={cats} color="green" />
  <LineUpCategoricalColumnDesc column="cat2" categories={cats} color="blue" />
  <LineUpNumberColumnDesc column="a" domain={[0, 10]} color="blue" />

  <LineUpRanking groupBy="cat" sortBy="a:desc">
    <LineUpSupportColumn type="*" />
    <LineUpColumn column="*" />
    <LineUpImposeColumn label="a+cat" column="a" categeoricalColumn="cat2" />
  </LineUpRanking>
</LineUp>;

CodePen

Result is same as the builder advanced example

Angular 6 Support (nglineup)

An Angular wrapper is located at nglineup.

Installation

npm install --save nglineup

Minimal Usage Example

app.module.ts:

import { BrowserModule } from '@angular/platform-browser';
import { NgModule } from '@angular/core';
import { LineUpModule } from '../lib/lineup.module';

import { AppComponent } from './app.component.1';

@NgModule({
  declarations: [
    AppComponent
  ],
  imports: [
    BrowserModule,
    LineUpModule
  ],
  providers: [],
  bootstrap: [AppComponent]
})
export class AppModule { }

app.component.ts:

import { Component } from '@angular/core';

@Component({
  selector: 'app-root',
  templateUrl: './app.component.html'
})
export class AppComponent {
  readonly data = <any[]>[];

  readonly cats = ['c1', 'c2', 'c3'];

  constructor() {
    const cats = this.cats;
    for (let i = 0; i < 100; ++i) {
      this.data.push({
        a: Math.random() * 10,
        d: 'Row ' + i,
        cat: cats[Math.floor(Math.random() * 3)],
        cat2: cats[Math.floor(Math.random() * 3)]
      });
    }
  }
}

app.component.html:

<lineup-lineup [data]="data"></lineup-lineup>

CodePen

Result is same as the builder minimal example

Advanced Usage Example

app.component.html:

<lineup-lineup [data]="data" [defaultRanking]="true" style="height: 800px;">
  <lineup-string-column-desc column="d" label="Label" [width]="100"></lineup-string-column-desc>
  <lineup-categorical-column-desc column="cat" [categories]="cats" color="green"></lineup-categorical-column-desc>
  <lineup-categorical-column-desc column="cat2" [categories]="cats" color="blue"></lineup-categorical-column-desc>
  <lineup-number-column-desc column="a" [domain]="[0, 10]" color="blue"></lineup-number-column-desc>

  <lineup-ranking groupBy="cat" sortBy="a:desc">
    <lineup-support-column type="*"></lineup-support-column>
    <lineup-column column="*"></lineup-column>
    <lineup-impose-column label="a+cat" column="a" categoricalColumn="cat2"></lineup-impose-column>
  </lineup-ranking>
</lineup-lineup>

CodePen

Result is same as the builder advanced example

Vue.js Support (vue-lineup)

A Vue.js wrapper is located at vue-lineup.

Installation

npm install --save vue-lineup

Minimal Usage Example

const cats = ['c1', 'c2', 'c3'];
const data = [];
for (let i = 0; i < 100; ++i) {
  data.push({
    a: Math.random() * 10,
    d: 'Row ' + i,
    cat: cats[Math.floor(Math.random() * 3)],
    cat2: cats[Math.floor(Math.random() * 3)],
  });
}

// enable plugin to register components
Vue.use(VueLineUp);

const app = new Vue({
  el: '#app',
  template: `<LineUp v-bind:data="data" />`,
  data: {
    cats,
    data
  }
});

CodePen

Result is same as the builder minimal example

Advanced Usage Example

const app = new Vue({
  el: '#app',
  template: `<LineUp v-bind:data="data" defaultRanking="true" style="height: 800px">
    <LineUpStringColumnDesc column="d" label="Label" v-bind:width="100" />
    <LineUpCategoricalColumnDesc column="cat" v-bind:categories="cats" color="green" />
    <LineUpCategoricalColumnDesc column="cat2" v-bind:categories="cats" color="blue" />
    <LineUpNumberColumnDesc column="a" v-bind:domain="[0, 10]" color="blue" />
    <LineUpRanking groupBy="cat" sortBy="a:desc">
      <LineUpSupportColumn type="*" />
      <LineUpColumn column="*" />
    </LineUpRanking>
  </LineUp>`,
  data: {
    cats,
    data
  }
});

CodePen

Result is same as the builder advanced example

Polymer Support (LineUp-Element)

A Polymer 2.0 web component wrapper is located at lineup-element.

Installation

bower install https://github.com/lineupjs/lineup-element
<link rel="import" href="bower_components/lineup-element/lineup-element.html">

Minimal Usage Example

// generate some data
const arr = [];
const cats = ['c1', 'c2', 'c3'];
for (let i = 0; i < 100; ++i) {
  arr.push({
    a: Math.random() * 10,
    d: 'Row ' + i,
    cat: cats[Math.floor(Math.random() * 3)],
    cat2: cats[Math.floor(Math.random() * 3)]
  })
}
conat data = { arr, cats };
<lineup-element data="[[data.arr]]"></lineup-element>

TODO CodePen

Result is same as the builder minimal example

Advanced Usage Example

// arr from before
<lineup-element data="[[data.arr]]" side-panel side-panel-collapsed default-ranking="true">
  <lineup-string-desc column="d" label="Label" width="100" ></lineup-string-desc>
  <lineup-categorical-desc column="cat" categories="[[cats]]" color="green" ></lineup-categorical-desc>
  <lineup-categorical-desc column="cat2" categories="[[cats]]" color="blue" ></lineup-categorical-desc>
  <lineup-number-desc column="a" domain="[0, 10]" color="blue" ></lineup-number-desc>
  <lineup-ranking group-by="cat" sort-by="a:desc">
    <lineup-support-column type="*" ></lineup-support-column>
    <lineup-column column="*" ></lineup-column>
  </lineup-ranking>
</lineup-element>

TODO CodePen

Result is same as the builder advanced example

R, RShiny, and R Markdown Support

A HTMLWidget wrapper for R is located at lineup_htmlwidget. It can be used within standalone R Shiny apps or R Markdown files. Integrated plotting does not work due to an outdated integrated Webkit version in RStudio. Crosstalk is supported for synching selections and filtering among widgets.

Installation

devtools::install_github("rstudio/crosstalk")
devtools::install_github("lineupjs/lineup_htmlwidget")
library(lineupjs)

Examples

lineup(iris)

iris output

Jupyter Widget (to be released)

A Jupyter Widget wrapper for Python is located at lineup_widget.

Installation

pip install -e git+https://github.com/lineupjs/lineup_widget.git#egg=lineup_widget
jupyter nbextension enable --py [--sys-prefix|--user|--system] lineup_widget

Or, if you use jupyterlab:

pip install -e git+https://github.com/lineupjs/lineup_widget.git#egg=lineup_widget
jupyter labextension install @jupyter-widgets/jupyterlab-manager

Examples

Launch Binder

import lineup_widget
import pandas as pd
import numpy as np

df = pd.DataFrame(np.random.randint(0,100,size=(100, 4)), columns=list('ABCD'))

w = lineup_widget.LineUpWidget(df)
w.on_selection_changed(lambda selection: print(selection))
w

simple usage

from __future__ import print_function
from ipywidgets import interact, interactive, interact_manual

def selection_changed(selection):
    return df.iloc[selection]

interact(selection_changed, selection=lineup_widget.LineUpWidget(df));

interact example

Observable HQ

A ObservableHQ wrapper is located at lineup-js-observable.

data = {
  const arr = [];
  const cats = ['c1', 'c2', 'c3'];
  for (let i = 0; i < 100; ++i) {
    arr.push({
      a: Math.random() * 10,
      d: 'Row ' + i,
      cat: cats[Math.floor(Math.random() * 3)],
      cat2: cats[Math.floor(Math.random() * 3)]
    })
  }
  return arr;
}
import { asLineUp } from '@sgratzl/lineup-js-observable-library'
viewof selection = asLineUp(arr)

ObservableHQ

Minimal Result

Advanced Usage Example

// arr from before
viewof selection = {
  const b = builder(data);
  b.column(
    LineUpJS.buildStringColumn('d')
      .label('Label')
      .width(100)
  )
    .column(LineUpJS.buildCategoricalColumn('cat', cats).color('green'))
    .column(LineUpJS.buildCategoricalColumn('cat2', cats).color('blue'))
    .column(LineUpJS.buildNumberColumn('a', [0, 10]).color('blue'));

  // and two rankings
  const ranking = LineUpJS.buildRanking()
    .supportTypes()
    .allColumns() // add all columns
    .impose('a+cat', 'a', 'cat2') // create composite column
    .groupBy('cat')
    .sortBy('a', 'desc');

  b.defaultRanking().ranking(ranking);
  return b.build();
}

ObservableHQ

Advanced Result

PowerBI Custom Visual (under development)

A PowerBI Visual wrapper is located at lineup_powerbi.

Installation

TODO

Examples

TODO

API Documentation

See API documentation and Develop API documentation

Demos

See Demos, Develop Demos, and R Demos

Related Publications

LineUp: Visual Analysis of Multi-Attribute Rankings Paper Paper Website

Samuel Gratzl, Alexander Lex, Nils Gehlenborg, Hanspeter Pfister, and Marc Streit
IEEE Transactions on Visualization and Computer Graphics (InfoVis '13), 19(12), pp. 2277–2286, doi:10.1109/TVCG.2013.173, 2013.

🏆 IEEE VIS InfoVis 2013 Best Paper Award

Taggle: Scalable Visualization of Tabular Data through Aggregation Paper Preprint Paper Website

Katarina Furmanova, Samuel Gratzl, Holger Stitz, Thomas Zichner, Miroslava Jaresova, Martin Ennemoser, Alexander Lex, and Marc Streit
Information Visualization, 19(2): 114-136, doi:10.1177/1473871619878085, 2019.

Dependencies

LineUp.js depends on

Development Dependencies

Webpack is used as build tool. LineUp itself is written in TypeScript and SASS.

Development Environment

Try the Gitpod online IDE and start coding instantly in your browser.

Gitpod Ready-to-Code

Otherwise follow the steps to setup a local development environment.

Installation

git clone https://github.com/lineupjs/lineupjs.git -b develop
cd lineupjs
npm install

Build distribution packages

npm run build

Run Linting

npm run lint

Run Unit Tests

npm test

Serve integrated webserver

npm start

Run E2E Tests

via cypress.io

Variant 1: with prebuilt LineUp

npm run compile
npm run build:prod
npm run cy:open

Variant 2: with webpack-dev-server

first shell:

npm start

second shell:

npm run cy:start

Link develop version

In order to use the library during development in another repository, one need to build and watch the library and produce development typings.

ln -s . <target_project>/node_modules/lineupjs
npm run compile:dev
npm run watch

The development typings are needed cause during production the are located at /src. That causes problems cause during compilation of a dependent project the Typescript compiler will first find the original TypeScript file e.g. config.ts before looking for config.d.ts, will complain that the library owner should deliver JavaScript files and won't compile the file. Thus the typings have to lie at a different location in this scenario.

Authors

  • Samuel Gratzl (@sgratzl)
  • Holger Stitz (@thinkh)
  • The Caleydo Team (@caleydo)
  • datavisyn GmbH (@datavisyn)

This repository was created as part of the The Caleydo Project.