Bi-directional Bar Chart

Introduction to Bi-directional Bar Chart

Bi-directional Bar Chart (also known as Positive-Negative Bar Chart) uses positive and negative bars to display numerical comparisons between categories. The categorical axis represents the dimensions to be compared, while the continuous axis represents the corresponding values. There are two cases: one where the positive and negative scales are completely symmetrical, and another where they are inversely symmetrical (i.e., opposites).

Other Names: Bi-Directional Bar,Bi-Directional Column

Composition of Bi-directional Bar Chart

Chart Type	Bi-directional Bar Chart
Suitable Data	List: one categorical field, two continuous fields
Function	Compare numerical values across categories
Data Mapping	Categorical field maps to position on categorical axis; one continuous field maps to length of positive bars; another continuous field maps to length of negative bars; positive/negative data can be distinguished by color or shape; categorical data can use color to enhance distinction
Recommended Data Count	No more than 30 items horizontally, 12 items vertically

Use Cases of Bi-directional Bar Chart

Suitable Use Cases

Example 1: Positive-Negative Category Comparison

The chart below simulates completion status of tasks by different departments in a company, with completed counts shown as positive bars and uncompleted counts as negative bars.

Department	Group	Completed	Uncompleted
Dept 0	Group0	37	9
…	…	…	…

import { Chart } from '@antv/g2';

const chart = new Chart({
  container: 'container',
});

const data = [
  { department: 'Dept 0', group: 'Group0', people: 37, type: 'completed' },
  { department: 'Dept 0', group: 'Group0', people: 9, type: 'uncompleted' },
  { department: 'Dept 0', group: 'Group1', people: 27, type: 'completed' },
  { department: 'Dept 0', group: 'Group1', people: 10, type: 'uncompleted' },
  { department: 'Dept 1', group: 'Group2', people: 37, type: 'completed' },
  { department: 'Dept 1', group: 'Group2', people: 19, type: 'uncompleted' },
  { department: 'Dept 1', group: 'Group3', people: 37, type: 'completed' },
  { department: 'Dept 1', group: 'Group3', people: 29, type: 'uncompleted' },
  { department: 'Dept 2', group: 'Group4', people: 20, type: 'completed' },
  { department: 'Dept 2', group: 'Group4', people: 2, type: 'uncompleted' },
  { department: 'Dept 2', group: 'Group5', people: 40, type: 'completed' },
  { department: 'Dept 2', group: 'Group5', people: 10, type: 'uncompleted' },
  { department: 'Dept 3', group: 'Group6', people: 25, type: 'completed' },
  { department: 'Dept 3', group: 'Group6', people: 3, type: 'uncompleted' },
  { department: 'Dept 3', group: 'Group7', people: 55, type: 'completed' },
  { department: 'Dept 3', group: 'Group7', people: 8, type: 'uncompleted' },
];

const range = ['#7593ed', '#95e3b0', '#6c7893', '#e7c450', '#7460eb'];

chart.options({
  type: 'interval',
  coordinate: { transform: [{ type: 'transpose' }] },
  autoFit: true,
  data: data,
  encode: {
    x: 'group',
    y: (d) => (d.type === 'completed' ? d.people : -d.people),
    color: 'department',
  },

  scale: {
    x: { padding: 0.5 },
    color: {
      type: 'ordinal',
      range,
    },
  },
  axis: {
    x: { title: '' },
    y: {
      labelFormatter: (d) => {
        return Math.abs(d);
      },
    },
  },
  style: {
    fill: ({ type }, i, data) => {
      if (type === 'uncompleted') {
        return 'transparent';
      }
    },
    stroke: (d, i, data, ...re) => {
      const { type } = d;
      if (type === 'uncompleted') {
        return range[i / 2];
      }
    },
    lineWidth: 2,
  },
  tooltip: {
    title: (d) => {
      return `${d.department}-${d.group}`;
    },
    items: [
      (d, i, data, column) => ({
        name: d.type === 'completed' ? 'Completed' : 'Uncompleted',
        value: d.people,
      }),
    ],
  },
});

chart.render();

Explanation:

Used color to distinguish different departments
Used shape (hollow vs solid) to distinguish completed vs uncompleted counts
Used position to distinguish different groups

Unsuitable Use Cases

Bi-directional bar charts are not suitable for data without opposite meanings

Extensions of Bi-directional Bar Chart

Example 1: Stacked Bi-directional Bar Chart

The chart below shows statistics of agreement levels for different types of people on a certain question, divided into five levels: Strongly disagree, Disagree, Neither agree nor disagree, Agree and Strongly agree. The negative levels are shown as negative bars with different colors. Sample data

type	Strongly agree	Agree	Neither agree nor disagree	Disagree	Strongly disagree
Question 1	11.7	59.9	18.5	9.1	0.7
…	…	…	…	…	…

import { Chart } from '@antv/g2';

const chart = new Chart({
  container: 'container',
});

const types = [
  'Strongly disagree',
  'Disagree',
  'Neither agree nor disagree',
  'Agree',
  'Strongly agree',
];
const colors = ['#c30d24', '#f3a583', '#cccccc', '#94c6da', '#1770ab'];

chart.options({
  type: 'interval',
  coordinate: { transform: [{ type: 'transpose' }] },
  autoFit: true,
  data: {
    type: 'fetch',
    value:
      'https://gw.alipayobjects.com/os/bmw-prod/82c97016-0f99-433b-ab21-9ecf14244610.csv',
  },
  transform: [{ type: 'stackY' }],
  encode: {
    x: 'question',
    y: (d) =>
      d.type === 'Disagree' || d.type === 'Strongly disagree'
        ? -d.percentage
        : d.type === 'Neither agree nor disagree'
        ? -d.percentage / 2
        : +d.percentage,
    color: 'type',
  },

  scale: {
    x: { padding: 0.5 },
    color: { domain: types, range: colors },
  },
  axis: {
    x: { title: '' },
    y: {
      labelFormatter: (d) => {
        return Math.abs(d);
      },
    },
  },

  tooltip: {
    items: [
      (d, i, data, column) => ({
        name: d.type,
        value: d.percentage,
      }),
    ],
  },
});

chart.render();

Explanation:

Used color to distinguish different agreement levels
Used position to distinguish different questions

Comparing Bi-directional Bar Chart to Other Charts

Bi-directional Bar Chart vs Bar Chart

Bar charts can only plot on one axis, with each bar representing one data point
Bi-directional bar charts contain both positive and negative axis, with each bar potentially representing both positive and negative data