D3.js - Packs of circles to present hierarchical data

TIL

635

@100daysTIL

72

D3js

6

interpolation

2

Minh-Trung Nguyễn viết ngày 30/01/2018

D3.js - Packs of circles to present hierarchical data

Note1: I decided to challenge myself with a series of "Today I Learn" for 100 days. Non-stop. This is the TIL of day 08. One day one question one short answer. Please join with me. Proposed tag: #100daysTIL.

Note2: To make it more difficult, for every 7 posts in Vietnamese, there will be one in English.

I have just finished another chart today in D3.js, it's Circle Pack. Here are some notes before next move:

What is "Circle Pack"

• Circle Pack chart belongs to a group of charts that help visualizing hierarchical data.
• We can think of hierarchical data as a tree:
  • the "top" node is called "root", there is only one "root".
  • Passing through all middle-level nodes (which have both parents and children), we arrive at the lowest level of the tree called "leaves".
  • Leaves have only parents, no children attached.
• To present hierarchical data in charts, we have many options:
  • Tree
  • Treemap
  • Cluster
  • Dendrogram
  • Circle Pack
  • Partition
• The name "Circle Pack" might be explained via its visualization: each level of data is gathered in one pack. Each pack is illustrated as a circle.

An example of Circle Pack chart:

To draw a "Circle Pack" in D3.js

We need to go through 2 big steps:

Step 1: Prepare the hierarchical data.

Here is the sample of a hierarchical data. In case of having other kind of data (CSV for example), it's required to convert it into the correct format with method d3.hierarchy().

Step 2: Draw the chart "Circle Pack"

• Setup the pack layout with d3.pack(). Information such as size, padding must be entered at this step.
• Pass the hierarchical data to the pack layout.
• Start the data join, for each data-embedded element inside the pack layout:
  • draw a circle
  • insert label

To make the "Circle Pack" zoomable

This is one technique. There might have other techniques that I don't know yet.

• Disable the pointer-events triggers for root and all leaves.
• Insert event listener "onclick" into each middle-level nodes.
• If node is "clicked", run the function zoom.
• Until its turn, zoom will call:
  • d3.transition() which returns the interpolator for calculating the tweens between current view and targeted view. This helps to run smoothly the animation from current view and targeted view.
  • the interpolator will be passed into function zoomTo()
  • function zoomTo() is to calibrate dimensions of all circles and text of the targeted view to fit with current svg' size.
  • the trick here is to keep varible view at global scope, so it can be updated frequently by function zoomTo().

Sample code can be found here: https://bl.ocks.org/mbostock/7607535

Conclusion:

• The logic of code is fairely straightforward, but it needs more time for digging further to really understand d3.transition() and d3.interpolate inside out.
• To check if the feature "zoom" could be wrote in a different way.

References:

• D3js Hierarchy

ngminhtrung 30-01-2018