Data visualization has continued to be a principal component of my work across many projects in different domains. Often I will happily hook up existing visualizations from various popular front-end packages, but just as often I end up writing new visualizations to target more specific needs. For this purpose, my primary starting point for many years was typically D3, but its somewhat less-than-desirable integration with React projects (though I have created various React experiments to integrate it into the DOM / virtual DOM in different ways) increasingly led me to different methods and tools.

When exploring a dataset for my own purposes and examining its contents, I most often begin instead with Jupyter notebooks, which even offer a perfectly acceptable space to embed a bit of exploratory D3 after slicing and plotting.

I once wrote an entire animated mini-film as a gift to my wife coded entirely in vanilla D3 code, using a hand-calculated camera / parallax / zooming scene system, photo assets of her artwork with many svg filters and distortions applied, and more tricks. D3 is a strange choice for the job, but it made for an enjoyable experiment. Watch the first 30 seconds here.