Kirk Goldsberry, a geographer from Penn State, was interested in finding ways to visually depict data about movement through space and time. He also was a big basketball fan who played all his life. In 2011, Goldsberry had the idea of mapping the dynamic ebb and flow of the game based on recently developed baseball statistical analytics. Using data scraped from ESPN basketball statistics web pages containing shot statistics, he eventually compiled spatial coordinates for more than 700,000 successful shots taken from 2006 to 2011. The final results were mapped as color-coded, square-foot pixels across the court (http://www.wired.com/2014/10/faster-higher-stronger/#slide-4):
This work led to a presentation made at 2012 Sloan Sports Analytics Conference, an annual gathering of statisticians and coaches at MIT. NBA coaches saw the value of the spatial patterns generated. A company called Stats teamed with Goldsberry to build a 3 camera-based system to track players and provided much more detailed information. In September 2013, Stats sold SportVU to the NBA for $100,000 per arena. With the spatial patterns generated by this system, spatial data analytics at the basketball court (micro) scale is now possible.
Now the NBA has the statistics available at http://stats.nba.com with their own version of shot maps, plus many detailed tables of every aspect of the game by player and team:
While there isn’t a public API provided by the NBA to access the statistics, there are web-scraping APIs available to retrieve the data programmatically, a much better alternative to the manual method Kirk Goldsberry used with ESPN.com. Micro-mapping articles and code samples are now available using the NBA statistics: The following map shows a Python language mapping utility written by Savvas Tjortjoglou (http://savvastjortjoglou.com/nba-shot-sharts.html):
This application takes mapping to a micro scale measured in feet, within a room, and not miles across counties, states, or countries. As a result, new mapping applications are feasible to show spatial patterns as long as the data is available!