Table of Contents
since I started learning D3.js as part of the 100 Days of Code project the number one question I’ve gotten is: how are you learning D3? I’ve had lots of people that want to know what learning resources I’m using, what my process is, and what I suggest for how to start on D3. I don’t think I’m any sort of expert, and I certainly don’t claim to have the ultimate learning process, but now that I’m halfway through the project I decided to share what I’ve learned. By the way, if you’re looking for a list of links, Nadieh Bremer has a wonderful list.
Disclaimer: This is merely an account of my own personal experience learning D3. It does not constitute professional advice and has not been vetted or researched at all. This is certainly not the right way to learn D3; it worked for me, and it may work for you, but there is nothing objectively correct about it.
Before I started my official 100 days, I started reading Scott Murray’s book “Interactive Data Visualization for the Web”. I read about half the book and it gave me a great primer on the lingo and common patterns in D3. One downside is that it’s written for D3v4, but now we’re on to D3v5, so there’s some key differences that aren’t covered in the book.
Observe and learn
Here’s what I wish I had known when I started using Observable:
- Observable cells are reactive and run in topological order. That means that anytime one cell updates, it will trigger a re-run of any cells that depend on it. This means you really need to think about how you code things like the D3
- Cells that are longer statements need curly braces and always need
returnstatements. Still today I forget to
returnthings when working in Observable.
Having introduced Observable, here’s my typical process for learning D3. I decide on a new chart I want to make, let’s say I want to make a Sankey diagram. I search Observable for a Sankey diagram example and give it a read. Then I search for some blogs on how Sankey’s work in D3, with special attention on how to prepare my data for the chart. As with ggplot, having your data in the right format is often half the battle, so it’s important to pay close attention to data preparation. Then I will go back to Observable, spend some time getting my data in the right format, and then try to emulate the chart, using the example notebook as a guide. Finally, I’ll play around with various parameters, colors, and other options to personalize the chart. There’re a few steps along this process where things can go wrong, here’s how to get the most out of it:
- Beware version differences. A lot of tutorials I find on blogs are written in v4 or v3 of D3, but a lot of the content on Observable is written in v5. The differences between versions are not difficult to parse for longtime users, but for beginners it can really throw you off. A lot of code is portable between versions, but it’s something to be careful about.
- As stated above, be very careful in trying to port code you find on blogs or Blocks to Observable, always keep in mind the Observable quirks.
- Don’t just copy-and-paste code. I see no problem in copying-and-pasting in certain cases, but it’s not a great way to learn, or at least it’s not a great way to learn if it’s all that you do. Even if you’re going to entirely reproduce an example, I think there’s some benefit in typing it out yourself. And even more important than typing the code, is making sure you understand it. Sometimes I’ll copy examples to get something working, but it’s so so important to go back and walk through the code line by line to make sure you really understand what each line is doing. Observable is perfect for this because you can tweak parameters, change variables, comment things out, and see what happens.
Some final tips
I’ve talked a lot about what not to do, so here’s some things I think you should do:
- Work with the same dataset a lot. I think it’s really helpful to find one dataset (preferably something large that has lots of options for plotting) and work with that throughout your learning process. You become more familiar with the data as you go along, and it helps reduce cognitive fatigue to have to worry about data cleaning and stuff when you’re also trying to learn D3.
- Ask for help. People in the D3 community are very nice, and I am very grateful to all of the lovely people that have helped me along the way. Join the Data Visualization Society Slack, or the D3 Slack, and if you’re stuck, just ask, more than likely someone will fix your code and tell you what was wrong.