Data Is Plural

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

2016.08.10 edition

Pretrial inmates, slave populations, hospital ratings, rocks, and heartbeats.

Pretrial inmates. Connecticut has begun publishing a daily census of every inmate held in jail while awaiting trial. Starting July 1, the database contains one row per inmate per day; each row includes basic demographic data (age, gender, race), as well as the inmate’s bond amount, main offense, and jail location. Read more at: The New Haven Independent and TrendCT. Question: This release seems unprecedented; does any other state or country publish such detailed data on pretrial inmates? [h/t Camille Seaberry]

U.S. slave populations, 1790–1860. For more than a century, the U.S. Census collected slave population figures. An assistant professor at George Mason University has aggregated that data, and mapped it. He cautions: “Treat the Census numbers skeptically: even in the best of circumstances the Census undercounts the population.” Previously: New Orleans slave sales in the December 30 edition; slave ship voyages in the January 20 edition.

Hospital ratings. The Centers for Medicare & Medicaid Services evaluates hospitals on dozens of measures — relating to safety, timeliness of care, patient satisfaction, and more — and publishes the results online as the “Hospital Compare” dataset. The dataset also includes an overall score, which distills each hospital’s results into a single five-star rating. If you don’t want to download the data, you can explore the results online. [h/t Drew Ivan]

Rocks. Macrostrat.org provides data and maps on thousands of geologic formations around the world. The database currently includes 1,474 “regional columns,” 33,903 “rock units,” and 1,750,044 “geologic map polygons.” You can also explore the data through the University of Minnesota’s “Flyover Country” iOS and Android apps. [h/t Grant J. Smith]

Heartbeats. PhysioNet has published sound and data files for more than 3,000 heart recordings (a.k.a. phonocardiograms). The files support PhysioNet’s 2016 contest, which seeks algorithms that can detect abnormal heart sounds. [h/t Joe Isaacson]