[rchaud]

Is it just me or is D3 absurdly complicated for making a simple chart? I look through that tutorial and I ask myself why I'd go to the trouble of coding a chart from scratch when I've presumably already spent time analyzing and cleaning a dataset and am looking to make it presentable to an audience.

If I had to go with a custom design, I'd use HighCharts (free for non-commercial use) or ChartsJS (free), which already have these basic functions (interactive hover tooltips with supplementary data) built in. Just declare your data variable and specify your chart type, and that's it. You can also modify the CSS to change the look and feel. Non-coders could build a dash in Google Data Studio, PowerBI or Tableau and export it for embedding on the web.

Don't get me wrong, I've seen Mike Bostock's D3 gallery and yes, it's ideal for making incredibly elaborate, artistic data visualizations. But a line chart should be simple.

[Isamu]

Yes, if all you wanted was a simple chart I wouldn't take the time to learn D3 unless I was motivated to learn it for other reasons. There are simpler solutions for simple charts.

I have used D3 for complicated visualizations, for example an orthographic view of stacked floor plans, you click on a floor and it slides out, zooms and re-orients itself and displays the time- and location-based data in different colored areas around the floor.

Complicated things like that is where D3 shines I think.

[wattenberger]

Exactly! I should probably have a section (or at least a link to a resource) that talks about _why_ you would want to learn this.

I have an image in the book detailing the spectrum of data viz tools on the web. At one end are the quick-pickup tools that aren't customizeable, and at the other end you have.. d3. Which isn't really a framework so much as a collection of tools that can help with making web charts.

What I probably failed to communicate in this post is that there are tons of opportunities to help communicate data with the web. Tooltips are just the most common interaction, but you could easily envision using scroll as a trigger, or adding a way to "zoom in" on parts of a chart.

The learning curve is definitely steeper, but once you have the skills, the world is really your oyster for visualizing data online. The pudding (https://pudding.cool/) is a great showcase of possibilities, or even uncommon chart types like at https://wattenberger.com/fishing

[stephen_g]

When the charting library gives you exactly what you want (or close enough), sure. Probably for 95% of what most people need to do that's enough.

But as soon as you need to deviate or add extra functionality from what your chart library gives you, it's often about as much work to just do the chart from scratch in D3 than to dig into the internals of ChartJS or whatever to add in what you want.

Also, while it does have a steep learning curve, once you get the hang of it it's not that hard.