Using the same National Weather Service data that powers his live-ish wind map of Earth, Cameron Beccario put together a time-lapse for all of 2018. Watch it on full-screen in its 4k glory.

[arve url=”https://www.youtube.com/watch?v=obsw9qiBnjo” /]

As the shutdown continues, 800,000 government workers wait for something to happen. The New York Times uses others industries for scale. Ugh.

This vintage recreation by graphic designer Scott Reinhard fills all the right checkboxes for me.

Denise Lu for The New York Times provides a quick overview of the proposed border wall and its progress. Scroll for zeros.

The end of a year is always a good time to look back at past work, because the day-to-day can sometimes feel like an endless churn. There’s also just no way to remember everything, and because of the volume, there’s really no way you can see everything that everyone made. Hence, lists. And Maarten Lambrechts compiled a list of all the visualization lists for 2018 for your perusal.

I am partial to the lists from The New York Times, The Washington Post (not listed), and FiveThirtyEight.

Nick Barrowman on the myth of raw data:

Assumptions inevitably find their way into the data and color the conclusions drawn from it. Moreover, they reflect the beliefs of those who collect the data. As economist Ronald Coase famously remarked, “If you torture the data enough, nature will always confess.” And journalist Lena Groeger, in a 2017 ProPublica story on the biases that visual designers inscribe into their work, soundly noted that “data doesn’t speak for itself — it echoes its collectors.”

So, when you work with data and make conclusions, you must consider everything that came before.

D3.js can be used for a lot of things, and for some people it’s too much to deal with.

Niklas Elmqvist provides a detailed guide for finding and a visualization PhD program:

Unless you have a specific reason to choose a specific university (such as a geographic one; maybe you can’t relocate), don’t start from the university you want to go to, but start with the faculty member you want to work with. This is where all that idle web surfing experience can come in useful: you need to become an expert in finding faculty members that have research interests that match your own, and the only way to do so is to trawl their websites and read their papers.

And then applying:

Now, having identified some possible advisors (and don’t just pick one; you never know whether you will be admitted and whether they have funding to hire new students), you should reach out to them. In other words, don’t just apply, but send them an email with plenty of time to spare before the application deadline. Attach your CV, outline your background, and provide some of the above-mentioned commentary on their work and why you are interested in it (i.e., the “hook”). If you have a portfolio or website, link to it. Remember, no form letters!

Useful information here. You might also want to get a sense of flexibility in the department. I was two years into my PhD in statistics until I decided I wanted to go the visualization direction, which was a big switch from my original intentions of statistics education. Focus and interests tend to shift after you learn more.

Once you get into a program, see also my survival guide for avoiding burnout and finishing.

The stock market is in a state. So finicky the past few months. Kate Rabinowitz and Leslie Shapiro for The Washington Post provide a view further into the past for more context to the recent flux. The stretching time axis as you scroll makes for an easy-to-follow visual cue.

Max Read for New York Magazine describes the fake-ness of internet through the metrics, the people, and the content:

Can we still trust the metrics? After the Inversion, what’s the point? Even when we put our faith in their accuracy, there’s something not quite real about them: My favorite statistic this year was Facebook’s claim that 75 million people watched at least a minute of Facebook Watch videos every day — though, as Facebook admitted, the 60 seconds in that one minute didn’t need to be watched consecutively. Real videos, real people, fake minutes.

I wonder how the fake-ness level online compares to fraud IRL.

While looking through this year’s projects, picking out my favorites, I couldn’t help but reminisce about the times when the internet used to feel so care-free. It was more relaxed.

These days, there’s too much going on in the world for the internet to relax. Or rather, more of the world happens online now. This year, I felt like if I was going to spend time working on a project or writing something, it had to help people see a different perspective or teach something. I couldn’t just do it out of personal interest.
Read More

Bloomberg charted voter turnout for the just past midterm elections, comparing 2018 against 2014. As you might expect, there are a lot of blue arrows pointed up and to the left. Turnout decreased in only two districts.

Matthew Conlen provides a short explainer of how kernel density estimation works. Nifty.

Kirk Goldsberry is back at ESPN. I put this here mainly because it’s nice to have the hexbin shot charts in the feed again.

The Wall Street Journal highlighted a disagreement between data and business at Netflix. Ultimately, the business side “won.” However, maybe that’s the wrong framing. Roger Peng describes the differences between analysis and the full truth:

There’s no evidence in the reporting that the content team didn’t believe the data or the analysis. It’s just that their fear of damaging a relationship with an actor overruled whatever desire they might have had to maximize clicks or views. The logic was probably along the lines of “We may take a hit in the short-run but we will benefit from this relationship in the long-run.” Whether that’s true or not is unclear, but it’s a tricky question to answer with data. It’s not even clear to me how you would formulate that question.

Data often pitches itself as the path to definitive answers, but most of the time it gives you possibilities and weighted suggestions. Follow blindly, and you end up with creepy, algorithmically-generated YouTube videos.

Descartes Labs used machine learning to identify all of the trees in the world where at least one-meter resolution satellite imagery is available. Tim Wallace with the maps:

The ability to map tree canopy at a such a high resolution in areas that can’t be easily reached on foot would be helpful for utility companies to pinpoint encroachment issues—or for municipalities to find possible trouble spots beyond their official tree census (if they even have one). But by zooming out to a city level, patterns in the tree canopy show off urban greenspace quirks. For example, unexpected tree deserts can be identified and neighborhoods that would most benefit from a surge of saplings revealed.

Jon Keegan scraped the playlist from the local radio station’s all-Christmas playlist for a few days. Then he looked at play counts and original composition dates:

Considering the year in which each song was written, my dataset spanned 484 years of published music. Of course, many of the older songs are considered “traditional” songs, without a clear writer or composer. One obvious thing about this genre is that it is rich with covers (performing a new version of someone else’s song). Of the 1,510 songs played over this period that I was examining, it turns out there are really only about 80 unique songs in the dataset. But from those 80 songs come lots of covers, medleys and live recordings.

Edward Tufte criticized R for not being able to do some things typographically. It came in a tweet and was likely misunderstood. Sort of. I got a clarification from the man himself.

Computers can generate faces that look real. What a time to be alive. As it becomes easier to do so, you can bet that the software will be used at some point for less innocent reasons. You should probably know how to tell the difference between fake and real. Kyle McDonald provides a guide to the telltale signs of AI-generated faces.