Adventures in data cleaning: Did the New York Times undercount risky San Francisco skyscrapers?

It looks like the New York Times may have undercounted the number of risky skyscrapers in downtown San Francisco, 48 instead of 39. It’s a seemingly small difference – 20 percent if you do the math – but if it’s significant if you consider how many people work in these large buildings. A June 15 story focused on steel moment buildings cited in a USGS report.

I made a quick map using the addresses from the NYT story, then I wanted to make one that included photos of the buildings. This time I went directly to the report, noticing that the first address wasn’t listed in the story, it seemed like a good idea to see if there were any more discrepancies.

TL;DR

To check how I got them:

  • The USGS report, starting on page 360, in .PDF
  • The .KML file I made, for more fact-checking and map making (pretty please send links to your maps or put them in the comments – I’m a casual mapper, using new tools and working quickly!)
  • Here are the additional nine addresses from the report that weren’t in the NYT story:
  1. The Mills Building, 221 Montgomery Street
  2. 225 Bush Street
  3. 140 Montgomery Street
  4. 120 Montgomery Street
  5. 45 Fremont Street
  6. 55 2nd Street
  7. 555 Mission Street
  8. 611 Folsom Street
  9. 680 Folsom Street

The clumsy adventure

To start, I downloaded the 454-page .PDF, then extracted five pages with the buildings listed by using the >Print>Pages>Save as .PDF function in Preview for Mac. Then I converted the .PDF to .CSV with Sejda. After that, it was time for Terminal to merge the extracted data from those pages into one file with the command:

cat *.csv >merged.csv

Still too messy to be useful without a lot of tedious cleanup:

So I tried the quickest and dirtiest way I know: copy the table from the .PDF into Word, then from Word (where it’s recognized as a table) copy it into Excel.

There are a couple hundred buildings listed, but the ones cited in the story are steel moment frames. Erected before a 1994 building code outlawed a flawed welding technique, they harbor particular risk in a quake of magnitude seven or higher.

From the USGS report: Steel moment frame listed as “Steel MF,” “Steel moment frame” and “MF.”

From there it was a question of sorting the buildings listed as “Steel MF,” noting that a couple are listed alternatively as “Steel moment frame” and one as simply as “MF.” Messy messy messy: also, totally typical. (There were also about 15 more listed as Steel MF in combination with some other reinforcement, since it would require more reporting to figure out if they’re as risky, these were left out.)

Then I checked the addresses against the story, added polygons for the nine new addresses to the previous uMap, downloaded it as a .KML file and started playing around in Google Maps.

The resulting map is a little disappointing. For starters, the polygons from uMap (which uses OpenStreetMap) don’t jibe that well with Google. As for the images – since the real a-ha if you live or work in San Francisco is how many of these buildings you’re in or around – I always forget how bad these are in the noob version of Google Maps. When you’re editing in the map, they are Polaroid-style pop-ups that resize whatever pic you throw in. The published version looks nothing like that and the overall effect with these building shots (all vertical) is horrific. Ugh. There’s no way to resize the window from this version of Google Maps – the alternatives are Google Fusion tables (which wouldn’t solve the problem here since AFAIK it works with points, not polygons) or  programming via the Google Maps API.

Why this happened

So how did the New York Times undercount the number of especially shaky high rises? Going on my experience with newsrooms (long) and with data (short but painful) my first guess is that the USGS mistakenly gave the Times an Excel or .CSV file that was different from what ended up in the final report.

The reporter knew there were enough buildings to warrant a story, somewhere around 40, the graphics person had the file, made the map and those numbers were plugged into the story and fact checked without going back to the published report.

Or there was some glitch between the formats – given how annoying the process of getting information from .PDF into anything – it’s easy enough. Data cleaning is the least interesting, most tedious part of any project. In this case, if I’m right, there are 20 percent more risky buildings than originally reported.

A quickie map of San Francisco’s earthquake prone skyscrapers


See full screen

See full screen – search for San Francisco if you see a world map.

The New York Times recently ran a story about San Francisco high rises – mostly downtown and South of Market – with steel frames that harbor particular risk in a quake of magnitude seven or higher. About 40 of these skyscrapers, erected before a 1994 building code outlawed a flawed welding technique, were cited in an April USGS report.

It’s one of those stories that could’ve used in interactive map at its core, but instead (it’s the news business, kid!) the map was a small, static graphic (see below) and the story ended with a list of the addresses.

Image courtesy NYT.

So here’s a simple map of those 39 steel moment-frame buildings. A few necessary caveats: this is the handiwork of a casual mapper trying out a new tool. I’ve been looking for a way to use OpenStreetMap to make personalized maps and spotted some earthquake maps from the Japanese OSM community with uMap, so it seemed worth a try. It was heavy going for a map made on the fly – the polygon tool was clunky and importing the list as a cleaned up .CSV wasn’t happening.

Still, a few things pop out: A few of these risky buildings are also near construction sites. In OSM, these are shown in sage green. (The light green represents parks.)

The struggle to use the uMap polygon tool is real. This is a closeup of 550 California Street, with a 19-story office building under construction nearby.

The Folsom Bay Tower will be a 39-story, 422-foot (129 m) residential skyscraper.

Park Tower at Transbay will have 43 stories, First & Mission’s Oceanwide Center features 636-foot-tall tower on Mission at First Street and a 910-foot-tall tower on the opposite corner on First Street.

And much like the reporter, shocked to discover the NYT offices are in one of these buildings, there were a few a-ha moments. A family member works in one and I’ve been inside at least a handful recently – an event at Autodesk, a movie at Embarcadero Center, a Wikimedia meetup, met a friend staying at the Marriott, emerged from the Montgomery Street Station in front of one three or four times, etc.

It’s an unscientific sample size of one (well, two if you count the reporter) but would wager that most people who live or work in San Francisco are around, if not inside, these buildings frequently.

Geographer maps San Francisco’s bike politics

Copenhagen has a lot more in common with San Francisco than most people think, says San Francisco State geography professor Jason Henderson.

While many look to the capital of Denmark as a Nordic idyll where the drin of bicycle bells outnumbers the blare of car horns, Henderson says it went through the same political fights to get there. “It’s not a magical unique place, actually, and that opens up the doors to possibility,” says Henderson, who spent a 2016 research sabbatical in Copenhagen and has a forthcoming book about the two cities.

Speaking at a recent Nerd Nite, Henderson gave some gears to grind as San Francisco heads into June 5 elections. Politics matter – how streets are configured, how much car ownership is taxed, how much space is allocated and protected for car parking and who decides these issues – and the daily habits of politicians matter, too.

“It’s important if we’re going to have not just a bicycle city but a truly sustainable transportation city,” he says. The problem? Few San Francisco politicians are really behind the bike as a method of transportation. Continue reading

Mapping UNESCO’s Intangible Cultural Heritage spots

Looking at the most recent UNESCO Representative List of the Intangible Cultural Heritage of Humanity it’s clear that these “elements,” as they’re called, are all over the map.

There’s painting, weaving, pizza making and spring rituals: but while they offer up videos, photos and text — there’s no actual map of these landmarks in sight.

Making that map shines a spotlight on why organizing data is crucial — and how every organization is a data trove and should be its own best data detective. Plotting visually can inform decision making and highlight patterns – inside trends to be worked into deeper groves or used to recalucate course. The list, according to UNESCO, is “made up of those intangible heritage elements that help demonstrate the diversity of this heritage and raise awareness about its importance.” Continue reading