"Sankey" Data Visualization with End-to-End Highlighting
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
asset
css
js
.gitignore
LICENSE
README.md
index.html

README.md

Sankey Diagram with End-to-End Highlighting


Live Demo

Link: QinMing.github.io/sankey

This diagram shows the survival statistics of the Titanic disaster. The thickness of ribbons (links) represents the number of people.

  • Move the mouse over nodes or links. The specific portion will be highlighted from end to end.
  • Double click on the diagram to turn off rich tooltips.
  • Drag the rectangles to move them.

This visualization was developed by Ming Qin, based on D3 Sankey plugin wrote by Mike Bostock (mike@ocks.org).

Features of the Visualization

1. Flow-based API and End-to-end highlighting

The original plugin take nodes and links as input, while this one has a different API. Instead of links, the input data contain flows, which have a single weight but multiple nodes in a chain. Here's an example of input data

{
    nodes: [
      {
        "name": "node0",
        "disp": "A",
      }, {
        "name": "node1",
        "disp": "B",
      }

      // ......

    ],

    flows: [
      {
        value: 50,
        thru: [ "node0", "node1"]
      }, {
        value: 30,
        thru: [ 0, 1, 2 ]
      }

      // ......

    ]
}

Compared to the link API, the flows are easier to construct because our raw data are in a multi-attribute table, where each row corresponds to a flow. Moreover, flows retain more information than links. It knows where the flow is coming from at any given place. Thus the end-to-end highlighting is made possible. Whenever the mouse hover over an element, there is a certain subset of flows going through it. Then some new links (called dlinks) are dynamically computed from this flow subset, and rendered in highlight. The endpoints of dlinks are positioned so as to make the highlighted flows look visually consistent. Though the positioning algorithm is coarse and can be improved.

The plugin is also very suitable for visualizing parallel sets. The demo shows the same data as parallel set visualization, but notice that their highlighting is very different.

2. Rich tooltips

The rich tooltip shows the subset of flows under your mouse. Double click on the diagram to switch between the rich and the simple modes.


Source code

Licensed under the MIT License. See LICENSE in the project root for terms