Sign up to get our latest work sent to you.
Interest in drones continues to increase in the United States. According to the research firm Gartner, drone unit sales are expected to grow this year by almost 40 percent. With news coming from the White House about a pilot program to expand commercial drone flights, the market seems primed for even more growth.
To help regulate an exploding industry, the Federal Aviation Administration (FAA) set up an unmanned aircraft registration system in December 2015. The FAA required anyone with a drone weighing more than 0.55 pounds to register before flying it outdoors for recreation. This requirement stayed in place until May 2017, when a federal court in Washington D.C. struck it down. Still, the data presented here includes nearly 820,000 drone registrations filed through February 2017.
Other researchers have looked at this data before in various ways. We've added an important methodological tweak in our approach. Not only do we view the data on a per capita basis -- adjusting for the population of each county -- but we also smooth the data with the Getis-Ord local statistic. This removes some of the variance in the raw data and makes "hot spots" on the map more obvious.
So where are the drone hobbyists located? According to our analysis, primarily in the middle of the country. Northern Colorado, eastern Montana, and the Dakotas over-index in drone registrations relative to the rest of the country. The western corner of Nevada and the Floridian peninsula also see plenty of drone action. Much of the South, meanwhile, remains a no fly zone.
As of February 2017, The Federal Aviation Administration’s registration data contains 42,327 records for a total of 661,179 drone registrants in the United States. For each record, we mapped the zip code to its U.S. county. In cases where a zip code spanned multiple counties, we assigned the drone registrations from that zip code to the county that encompassed the largest percentage of its addresses. We removed records with nonexistent zip codes.
To calculate drone registrations per capita, we used 2016 population estimates for each county courtesy of the U.S. Census Bureau.
We then computed the Getis-Ord local statistic for each county based on drone registrations per capita. Getis-Ord is a nearest neighbors technique that aims to identify "hot spots" in geospatial data. It assigns a z-score to every county in the data based not only on the given county's feature value, but also that of its k nearest neighbors. If the local sum for a county and its neighbors is very different from the expected local sum (i.e. the national average), a significant value results. For our analysis, we had the algorithm consider each county's 10 nearest neighbors in computing the Getis-Ord statistic.
Graphic created in Tableau.
Data comes from the Federal Aviation Administration's unmanned aircraft registration system. Full dataset is available here.
For a primer on regional smoothing with the Getis-Ord local statistic, check out The Pudding's guide. It provides a basic implementation of smoothing in the R programming language with line-by-line commentary.