Data Is Plural

... is a weekly newsletter of useful/curious datasets.

2017.08.30 edition

Flood maps, redlining maps, home price indices, website logos, and Game of Thrones personalities.

Flood maps. FEMA’s Flood Map Service Center publishes geospatial files that detail the agency’s flood risk assessments — both current and historical. The maps include flood zones, levee locations, “base flood elevations,” and more. Helpful: FEMA’s technical documentation. Related:Why Houston Isn’t Ready for Harvey,” published last week by ProPublica and The Texas Tribune; and “Hell and High Water,” the reporting team’s deep dive on Houston last year. Previously: The most comprehensive global dataset of cyclone paths (DIP 2017.04.19).

Redlining. The Mapping Inequality project has digitized more than 150 of the “security maps” produced by the Home Owners' Loan Corporation between 1935 and 1940. Together, the maps “offer a view of Depression-era America as developers, realtors, tax assessors, and surveyors saw it — a set of interlocking color-lines, racial groups, and environmental risks.” To download the data for a given map, click on the cloud icon in the top-right corner. Related: A new research paper, by economists at the Federal Reserve Bank of Chicago, uses the data to quantify redlining’s lasting effects. Also related: The New York Times’ summary of the data and research. [h/t Kendall Taggart]

Home price indices. The Federal Reserve Bank of St. Louis publishes S&P/Case-Shiller Home Price Index data, which measures changes in average home prices over time. The monthly-updated datasets — copyrighted, but free to download — are available at a national and metro-area level, and go back several decades.

Website logos. Favicons are the little square icons in your browser’s tabs, placed there by the websites you’ve loaded. Two recent projects attempted to collect these markers from the web’s million most-trafficked domains. One, by programmer Colin Morris, collected 360,000 favicons in July 2016. The second, by researchers at ETH Zurich, collected 548,00 favicons in April 2017. Semi-related: Morris’s “Finding bad flamingo drawings with recurrent neural networks”; the analysis uses Google’s 50-million-doodles data, featured in DIP 2017.05.04.

Game of Thrones characters, judged. Earlier this month, The New York Times asked readers to rate 50 of the show’s most recognizable characters along two dimensions: good ↔ evil, and ugly ↔ beautiful. They’ve received 190,000+ submissions. The results are accessible as two JSON files: one for the averages and another for the distributions.