BizWeekGraphics

Making your charts magenta and your numbers big.

[thelisaproject]

I’m in a train between New York and DC and I’m very tired and very happy. Today was the my last day in NYC and another great, great day. 

It started with a tour through the Wall Street Journal newsroom, where we heard from the data team about the last stories they have covered. Interesting to hear: their team deals with data, barely with data visualisations. Their final output is often a long article and some mainstream looking bar charts. The Wall Street Journal has a quite big data visualisation team, though. I’m not used to this divide between data visualisation and data journalism; between the Visuals team and the Data desk in a newsroom. I create data visualisations because I want to tell stories with data. Isn’t that exactly what the Data desk does? I guess the difference lays in the span of the projects: Data Vis projects seem to be shorter; Data desk projects tend to fall in the category “investigative journalism” and can easily take a few months to finish. I’m not 100% sure. I will investigate on that matter. Maybe with data.

The next stop was the Bloomberg tower. It was interesting and a little bit emotional to be back in the place where my love for data visualisations began; where I learned so much during my internship exactly 3 years ago. I met with Toph Tucker, who started working at the Bloomberg Businessweek a few months after I left the place, and he introduced me to many Visual Data people - it was great to meet them in real life.

And my last stop: ProPublica! Like the tour through the Wall Street Journal, that visit was organized by the Global Editors Network, who also organised the Unconference yesterday. Scott Klein and Sisi Wei gave a super interesting talk about their team’s stand in the newsroom. The News App Developer team seems to have an ideal attitude: All of them are journalists, designer and coder at the same time - or at least can be. And they see themselves as a “normal” newsroom section. Meaning, they not only support the other sections with their stories, but mostly create stories themselves (which reminds me a lot of how the NYT Graphics desk acts).

I found Scott’s opinion about coding especially interesting: He said that compared with writing and designing, coding is the easiest part – and he mentioned frameworks (like Angular.js) as an explanation. I’m not sure if I agree with that. I’m 70% convinced. I still feel like coding is something that gets valued extremely high by most people I know. But the reason for that might be mostly a potential fallacy: missing understanding and therefore too high respect for the field. Since I’m also a victim of that fallacy, I can’t even judge if it IS a fallacy.

It was great to see some of the other Unconference people again today at the tours. I feel weirdly close and connected to that “community of one day”, and I really hope I see as many of them again. I feel like I learned a lot from some and could have learned a lot from all of them. People, man. Awesome.

After saying goodbye to them, I wondered around the SoHo neighborhood. It was perfect weather, the sun was shining, and I enjoyed first Joe’s 2.75 Dollar pizza and then musicians in the subway. And I got more and more jealous of everybody who works and lives in New York city. My motivation to go back to DC is mostly limited to a) sleeping in a bed again (instead of on the couch of my fellow fellows) and b) to go back to work at NPR again. Yep, I am looking forward to that. But I wouldn’t mind having the NPR Visuals Team and my bed in New York. Or Berlin. Or any other really awesome city. Ah, I’m sure it’s happening in a parallel universe.

Input? 8

Output? 2

Learnings?

I should get to know the data desk at NPR.

Try to use as little statistical tools as possible to tell the story. As the WSJ Data team puts it: “We’re mostly happy when we’re counting, because then we don’t have to explain any statistical method to the editor…or to the reader.”

To have a solution in search of a problem is not a bad thing. It’s ok to have a hammer and look for nails. Just be honest about what really is a nail (and what just looks like one because you have a hammer).

Sometimes art is great to just bring people together and have a conversation. What we consumed is replicated in what we make. We make similar things as we consume. We have one taste, and sometimes we express it through consumption, and sometimes through creation

Questions?

Should I learn statistical methods for small datasets, or should I rather learn to deal with large datasets? (Get them, organize them, analyze them)

[risingtensions]

(via bizweek valentines on Vimeo)

my coworkers <3 <3 <3

[our-beginnings]

Allison McCann, Visual Journalist, FiveThirtyEight

Tell us a bit about who you are, and what you do now.

Hi I’m Allison! I grew up in San Diego but have somehow survived the last five years in New York. I’m a visual journalist at FiveThirtyEight, the data news site started by Nate Silver, where I research, report, write, design and develop stories.

We’re still a small newsroom, which is great because I can pretty easily pitch any idea in our weekly editorial meetings and get feedback and suggestions from everyone on where to get data, who to talk to, etc. I don’t have any specific mandate, so finding ideas is usually the hardest part! I try to read a lot, follow the news closely and keep up with specific interests— women’s soccer, pop music, abortion legislation (wow, my interests seem weird when they’re in a row like that)— and we also have an “Ideas” Slack channel where everyone is constantly throwing in good stuff. 

Keep reading

[risingtensions]

[bizweekgraphics]

“Now we wait”

[worldmapswithoutnewzealand]

Umm, okay

http://www.bloomberg.com/graphics/2015-paul-ford-what-is-code/#code-atlas

[bizweekgraphics]

Bizweekgraphics 1990–2001: Chartjunk.

Only 90s kids will remember these, if they happened to be abject dorks who read Businessweek. In 1990 came color and its attendant abuses, but if you embrace the "the phrase 'desktop publishing' was invented ten minutes ago" of it all it can be genuinely charming to have put a bar chart inside a hole, or made it out of shopping bags, or to have made a buggy whip sales line out of a buggy whip. A graph of printer company share price? Hell, put a printer behind it! 

The apt-hedded “Giving up” thing deserves special attention. Not only do the slices not add up to a whole, but they're not even slices of the right thing: the data is proportion of respondents who gave up an activity, not a proportion of any amount of time, which is what clocks traditionally show. Traditionally. Until the maverick graphics editors of a bygone era stood up to convention.

But the Investment Figures are clean enough (though the color doesn't add anything). The Merrill dude checkin out a 3D exposure surface (1992) is pretty dope. Fax survey looks fun (Dec. 1993), and nice diss in the footnotes: “Information Superhighway revenue to date: $0. Number of articles since 1992 on potential Superhighway profits: 38,410.”

The “Big brother” survey, in which a majority of Americans supported granting law enforcement agencies more power, appeared in the Oct. 22, 2001 issue. Ha ha we used to be soooo dumb back then! Right!

1929–1935: Depression. 1936–1939: Prewar. 1941–1945: War. 1948–1951: Postwar. — YOU ARE HERE —

[bizweekgraphics]

Bizweekgraphics 1948–1951: Postwar.

After fighting hard comes playing hard. Most of Europe is still literally rubble but it's all sunshine in the grand old USA. All that time grilling in a Levittown and driving a car that both runs on and weighs as much as a dead elephant works up a powerful thirst; a thirst for strong brown liquors, which you could obviously drink in an office, because how else could these terrible charts make it into our pages.

Production is up, factories are turning to pacific concerns, and enough folks are worried about food surpluses that the notion of a food shortage due to overpopulation is actually a contrarian #take. The income distribution is tidily bell-shaped-ish, with more Americans enjoying “The Good Life”. 

1929–1935: Depression. 1936–1939: Prewar. 1941–1945: War. — YOU ARE HERE — 1990–2001: Chartjunk.

(Our archives are a little spotty, so the next post will just skip ahead, uh, forty years or so...)

[bizweekgraphics]

Bizweekgraphics 1941–1945: War.

Sure, it left 73 million dead, but have we talked about the effect of World War II on #infographics?!?? Charts were labeled according to Allied and Axis nations, or pre- and post-war, or data was included only up till the last full peacetime year, or huge swaths of the world were denoted “cut off by war”. Different versions were drawn for different contingencies, like whether India and Ceylon (Sri Lanka) rubber deliveries would continue to reach the U.S. One chart pertains, according to footnote, to the “37 eastern states”, which was apparently a well-defined thing.

In the eye-catching but misleading oil production graph wrapped in a geological metaphor (Oct. 6, 1945), note how little oil the Middle East had yet produced, and how much was projected.

1929–1935: Depression. 1936–1939: Prewar. — YOU ARE HERE — 1948–1951: Postwar. 1990–2001: Chartjunk.

[bizweekgraphics]

Bizweekgraphics 1936–1939: Prewar.

America's still in the fog, but we're about to sell damn near everything hand-over-fist to the Brits. Strife on the continent? Old story. Good thing that's way over there.

Check out those sparkcharty bars and small multiples! Those pattern fills and polylinear time scales! Those swoopy arrows and cute icons! And that pie chart with overlapping, non-MECE groupings!

1929–1935: Depression. — YOU ARE HERE — 1941–1945: War. 1948–1951: Postwar. 1990–2001: Chartjunk.

[bizweekgraphics]

Bizweekgraphics 1929–1935: Depression.

Businessweek is born into the Roaring Twenties hangover. Ever listen to Hard Times? Yeah, that. Rickets and sawdust bread. America basically has cirrhosis. Things cost a nickel but that's a nickel you don't have. You march for a twelve-hour workday but a hired goon in a nice hat staves your head in with your own picket.

Why did those strikebreakers wear hard shoes? You couldn't ever be in your shirtsleeves, even for violence? But the charts: The Business Week started off right, with good clean graphics, useful illos, low cruft, high density, all boxes ticked. The tables were richer than what Excel has taught us to be accustomed to, the flowcharts were well-proportioned, and the maps tackled irregular regional partitions like a champ.

But do notice the not-to-scale bar chart of dimes, which proves that lying chartjunk is as old as...old as...this old ass magazine.

— YOU ARE HERE — 1936–1939: Prewar. 1941–1945: War. 1948–1951: Postwar. 1990–2001: Chartjunk.

[3eanuts]

September 26, 1985 — see The Complete Peanuts 1983-1986

[meanassmoses]

colorful, minutely detailed, ｕｈ-ｍａｚｉｎｇ nuclear reactor cutaways courtesy of Bibliodyssey and the UNM University Library. There’s over a hundred of them there and I zipped ‘em all into one file for you.

According to the Flickr you can order these bad boys from: wdal AT neimagazine DOT com hope that’s still true today.

[bizweekgraphics]

Process post: how to make a giant oil spill map the hard way

Pray listen, stranger, to a tale of noble deeds and bitter struggle, of fffffttt nope it's about me doing computershit for around three weeks. But I am proud of this graphic and interactive map on oil spills and it has a sort of Homeric-level story behind it. But only if the first thousand stanzas of the Odyssey were him trying to untangle rigging while yelling "Does anyone know how to sail?"

I pitched a bigass graphic timed with the 25th anniversary of the Exxon Valdez spill, which was both an ecological catastrophe and a watershed moment for petro-tragedy regulation. Recap: a tired third mate under the command of a drunk captain plowed an oil tanker into a reef, around 1/5 of its crude poured into a remote Alaskan sound, and a lot of wildlife death followed. Thousands of people hosed rocks and scrubbed birds, Exxon paid far out the butt, and the Oil Pollution Act of 1990 firmed up liability law for spills. 25 years later and there's residual crud stuck to beaches and fewer Orcas than there should be. So yeah, heavy hydrocarbons tend to linger. Try and keep them out of the biosphere.

I wanted a map showing all oil spills in U.S. waters since the Valdez. Step zero: find a dataset. I alighted on the National Response Center query tool, which is the pollution incident reporting clearinghouse run by the U.S. Coast Guard. It is, naturally, pretty dated-looking and also uses flash. A classic tale of news dev woe. I grabbed the yearly summary data files, found they were inexplicably wrapped in self-extracting .exes, took a round trip to my windows laptop, out came two dozen excel files with ~12 tabs each, which I then ran through some excel macros to split into 231 discrete CSVs covering the years 1990 to 2012. Phew.

From those I wanted to sift the incident date, the vector (ship, refinery, etc.), the stuff spilled (petroleum only), the spill amount and the location. Of course there was no documentation to help me sort the useful field names from the PTF_FLAG_Ys that made up the bulk of the column headers, but with a little bit of OpenRefine and a lot of emails, I figured I could likely sort that out.

So now I had data from 1990-2012 in a sorta-convenient format, but I had to use a python script to grab 2013-2014 pollution incidents from the query tool.

At least I would have done that but for being caught between the Scylla of website maintenance and the Charybdis of pokey FOIA approvals: the query tool went down, so I was not getting any more data. Wasn't too awful a fate since the editors okayed using data up to 2012.

But then I took a closer look at the data and noticed how locations were stored. Addresses. Cities. States. Oh lord, geocoding. I'd need to geocode an apocalyptic number of incidents, sure to overwhelm whatever API I could throw at it. A pall fell over the project. Mostly because I'm bad at thinking things through.

NICAR 2014 interrupted my panic, the news-Calypso to break up my hideous journey. I especially enjoyed David Fallis' talk on tracking guns and the four "intro to SQL" classes I had to camp out for (note to NICAR 2015: overflow rooms for the hands-on seshes would help). SQL, SQL...yeah...this gives me an idear.

As with most situations involving me and editorial graphics, fortune intervened to cut the knot. Hallelujah: there was another Coast Guard database that I could use. And it had distinct advantages:

It tracks chemical and oil spills specifically. Incidents of that sort are included in the National Response Center data set, but it's more of a giant pollution clearinghouse and thus cluttered with wacky stuff. Exhibit A: someone called the government to tell them a mercury thermometer broke.

Instead of locating the incidents by address, they have latitude and longitude columns! No desperate shitfights with geocoding confidence levels! Just beautiful decimals! Deliverance! Praise to William Harrison!

Now I had more data, in a better format, with better documentation. By pure luck. Not a good strategy but one can't argue with results. Anyway I knew I had to face the database music this time: Excel would choke and die on the 20mb TSVs I had to deal with. Again, lucky me, I'm blessed enough to have Toph Tucker at my elbow, a true Phaeacian who could be relied on to provide deep sighs and plenty of "you forgot the semicolon again"s.

I loaded my tab-delimited text files into a MySQL database using SequelPro (great product) and got to querying. Being a neophyte, this naturally took a while, so we'll skim over my troubles here. With Toph's help I was able to format queries to retrieve oil spills that a) actually involved petroleum products b) involved them in some non-picayune quantity and c) had lat/longs associated with them.

After all sorts of false starts and late nights, I ended up with a spreadsheet that included all oil spills from ships, facilities, pipelines and random dumpage, of petroleum products, of 1,000 gallons or more, from 1973 through May 2013. That's a winnowing of ~600,000 rows down to ~4,300.

Time to get plotting, right? Not so fast: while most of the rows had decimal lat/longs, some had this N25456 E055186 nonsense. C'mon son. Some Excel magic remedied that:

And I had my final output, looking damn fine if I do say so myself.

CartoDB was used for the public-facing interactive map, but also proved invaluable at this stage because it's so easy to prototype with. I could upload my master spreadsheet and within seconds see where the clusters were, what the data looked like as scaled circles, if any lat/longs ended up in the middle of Greenland, etc. Anything that minimizes the time spent futzing in MAPublisher is good in my book, a regular Aeolian wind bag (the Odyssey? Eh? Cute huh?).

The rest was relatively easy window dressing: major oil ports and pipelines came from the EIA, major commercial waterways came from the U.S. Army Corps of Engineers. The usual MAPublisher nightmares ensued, but I was on familiar territory for once:

The map's foundation was poured, so it was time to sift arcane spill remediation PDFs and conduct some interviews. My mother always told me "journalists make phone calls" so I rang up Dr. Jeff Short and the nice folks at the Exxon Valdez Oil Spill Trustee Council.

After the callouts were written and bones of the map were set, it was time to recruit the best art directors on the continent: Cynthia Hoffman and Chris Nosenzo. To demonstrate their talent I placed the first draft next to the final.

From assey to classy in only 30 drafts. I lost the geographic literalism that I began with, which bummed me out since I get real sweaty whenever I get to go all National Geographic carto on a project, but in the end the reader won out. Cindy and Chris tuned up the colors, fixed my type hierarchy, got some cuter illos, and a two-page spread was born. Honey I'm home clean your room Telemachus har har har har.

→→→The final product, loud and proud←←←

Next came the interactive map, which thanks to CartoDB, was mostly finished! CartoCSS made it trivial to match the magazine's color and line weight styles, but the combination of tiles and vector symbol scaling did not sit well with CartoDB:

No matter what I did, the Deepwater Horizon circle was too damn big to render properly. And to get the jump on any pedants, yes I remembered to do the square root thing. Anyway Cindy had the idea of saying "screw it" to the circle scaling, which looks better anyway:

Et voila. As a beautiful coda, Toph made a mobile-optimized teaser. I couldn't begin to tell you how he did it. Go ask him.

Until the next graphics nightmare, —Evan Applegate • on twitter • on tumblr

ＯＮＥ  ＭＯＲＥ  ＴＨＩＮＧ

Oh so you think you could have done better? Is that right, punk?? Well be my guest, make a better map! Seriously please do, Businessweek is hiring an editorial cartographer. Come map for us and spare my poor editors and art directors the goofery detailed above.

[bizweekgraphics]

★☆★☆ Bizweekgraphics is hiring! ★☆★☆

The Businessweek graphics desk makes neat charts, maps, comics and explainers for both print and web, with equal enthusiasm for daily, weekly and long-term projects. As part of the award-winning Businessweek design department we emphasize original thinking and close work between designers and devs to try new things.

We're currently expanding and looking for talent. If you have a nose for reporting, know that "the numbers don't lie" is itself a lie and the phrase "data viz" makes you cringe, then talk to us. You'll get a lot of creative freedom, a competitive salary, and the chance to shape the graphics presence of an extremely good outlet. Check out the postings below and send a cover paragraph, resume, and a few projects you did/are working on/want to work on to: bizweekgraphics [at] gmail dot com.

P.S. We're in lovely midtown Manhattan. Very original, I know.

News Developer We’re looking for a full-stack web developer to help us build dynamic graphics, news apps, tools and infrastructure. This is a job for someone who wants to grab readers' attention and get them to understand the more complex aspects of business, finance and technology.

JavaScript and d3js are central and critical, but you’ll use whatever needed to get the job done…which is going to be more than JavaScript and d3js. Something server-side, some good MV* framework, and your database of choice. A strong data sciencey or statistical background helps. Be warned/piqued: you won't be sliding into a hierarchy of devs. We're building the airplane during an Immelmann turn, so you'll necessarily be independent and self-directing. If you're not afraid to be the nucleus of our dev team, then come all, ye ambitious!

Data Journalist Do you dream of mastering VLOOKUP or scraping data off assey government websites? Have an itchy FOIA finger? Have a mint collection of NICAR shirts? Bubbling over with data-driven story ideas that span a variety of time scales, i.e. a daily project, a weekly, and a long-form slog? Have you ever started a project, thought "I'd need to learn ____ to even start this," and then learned that thing? 

Then we'd like to hear from you. Don’t worry if you're short on design skills (we can teach you that) but you need to be mega-curious and have little fear of phone calls. A background in statistics, math, or engineering would help, but at minimum you need a familiarity and sunny disposition toward numbers. You'll be working with Python, R, d3js, Excel, Illustrator and InDesign to get your stories and make them presentable.

Editorial Cartographer Of course you like maps. But to map for us, you need to be able to:

Know which lines to keep, which highway shields to place, and what kind of hierarchy a map needs to be legible

Know how to dig out raster and vector datasets on tight deadlines, e.g. find that one shapefile in a haystack of awful county websites

Cobble together an interactive map with Leaflet.js

Do a bunch of joins and queries with postGIS

Make a hillshade, then make the rest of the map look just as cute

Whip out a competent lil' static locator-style map within 20 minutes when news breaks

Come up with a rickety geocoding workaround for when one has too many addresses and not enough API keys

Pick up the phone and call whoever maintains the file you're poking around when, inevitably, you can't decipher a field name

You should know: the useful parts of Creative Suite, ArcGIS/QGIS, PostgreSQL + postGIS, familiarity with GDAL, Python, and d3js required. MAPublisher a plus.

Page through our tumblr and flickr to see examples of our carto, which is super thematic but completely terrain-less even when it could really use some. Hoping you could change that. Important to note: we're not looking for just another pair of hands. You will be the chief cartographer, the grand mapping poobah, with all attendant heaviness of that crown: you will have pretty much no one to turn to for answers. But if you like leeway, you got it buddy! So if you're all about figuring stuff out for yourself, rube-goldbering solutions together on deadline, and have an editorial sorta mind, give us an email.

[bizweekgraphics]

A better Pi Day (if you ignore Kepler)

“Pi Day” is borderline numerology. This has been bugging me, so I was glad to see Vi Hart's great video on the topic today.

Even assuming π is worth celebrating (which I think is reasonable), one problem is that, by our apparent collective folk methodology, Pi Day falls on very different days depending on which base you use to represent it:

(For the record, @cornfact's theory is that, like Hallmark "invented" Valentine's Day, Pi Day is a crass creation of Big Pie.)

But there might be a better day to celebrate something more substantial, less arbitrary, and closely related.

Earth's orbital period is 365.25636 days. Divide by 2π (or τ) and we find that we have subtended an angle with respect to the sun of one radian after 58.132355 days. So, at 3:10 a.m. on February 28, we have traveled along an arc as long as we are distant from the sun.

This annual event is invariant with respect to digital base, as well as with respect to your π/τ opinions (if any), and it gets at the actual point of π, which is… the relationship of radius to circumference.

(Except π shows up somewhat more often than that, so maybe there's something more fundamental? The circumference is just the set of equidistant points, where the distance is the radius, so it's, like, the ratio of the measure of the equidistant set to their common distance to a distinguished point? I don't know, I never studied Lebesgue measure. How exactly does the periodicity get into that? And multiplication by imaginaries is basically rotation, so the complex exponential function ends up showing basically the same periodicity as relates squares and circles… which seems curious, but only because I don't understand it.)

But wait! I have quietly been assuming the Earth's orbit is circular! (It's elliptical.) While trying to make the math half-respectable, I have terribly disrespected astronomy and physics.

Never celebrate anything.

Just kidding. Briefly, let's see… we sweep out equal areas in equal times, so when we're farther, we must be moving more slowly, and farther plus slower unambiguously means lower angular velocity, so the time to subtend one radian varies. Moreover, the tidy equality of radius and arc-subtending-one-radian is broken, as the perimeter of an ellipse is, uh, non-trivial.

So I don't know! What else have I got wrong?

I'll get back to work now.

—@tophtucker

[bizweekgraphics]

Do you dream of mastering the VLOOKUP function in Excel or scraping data off pesky government websites? How about learning how to turn words into charts, maps, lines, bars or scatter plots? Good news! We're looking for a graphics intern to join our data-wrangling team.

Don't worry if you aren't a designer (we can teach you that!) but ideal interns will have a background in statistics, math, engineering or, at a minimum, a no-fear attitude toward spreadsheets and numbers. Interns will learn how to find data, explore it, understand it, visualize and design it for both print and the web through programs such as your brain, Excel, Illustrator, InDesign and even a little Python, R and d3js. We can pay interns $12.50 for up to 20 hours per week -- after that, you're on your own! :) If you're interested, send three data-driven pitches to Cynthia Hoffman ([email protected]) and Allison McCann ([email protected]) and we'll see if you've got what it takes to to make chart magic. 

[bizweekgraphics]

The tortuous road to a McDonald's infographic. Spoiler: it's a table.

McDonald's decision to include cucumbers in the McWraps was a BIG DEAL. Space at their restaurants is scarce, so new menu items tend to be ever more complex assemblies of a limited number of basic ingredients.

Connecting all items by the ingredients they share begins to reveal the groups and patterns sitting within the system. And it's a beautiful system, one whose complexity surely reveals the involvement of some cognizant higher being.

The complicated subway map scenario. The bacon and cheese quarter pounder is Times Square, and reduced fat vanilla ice cream is the Staten Island Railway.

What if we remove the breakfast items and side dishes and start all the ingredients at the same place? Sure, we lose the beautiful little breakfast-lunch crossovers like the tortilla's foray into breakfast or liquid margarine's one lunch/dinner appearance. But it would look less complicated.

The last time I was at a McDonald's I had a hash brown, seconds removed from the deep frier. It was delicious - crisp, oily, easily the match of any tempura or fried frogs legs you could name. And this isn't less complicated at all.

OK, fine. We'll scrap it all and go for something people might *understand* and *enjoy*. How does one draw habanero ranch sauce?

The McDonald's itself was depressing so I ate my hash brown in the car. And I have a depressing car.

—Dorothy