The Real-Time Rat Tracking Map
Most transit maps show where people travel. RATFLOW NYC shows where the city's other commuters roam.
RATFLOW NYC is a fascinating "data-art" project that visualizes New York City's estimated rat population using public data, probabilistic modelling, and a healthy dose of theatrical flair.
Built using NYC Open Data, RATFLOW combines rodent-related 311 complaints with the locations of subway stations to create a city-wide rat-density model. The result is an animated map that is a hybrid of a transit dashboard, a rat surveillance system and rodent wildlife simulation.
Rats do not have smartphones. Therefore RATFLOW is not actually mapping real rats. As the site itself explains, there is no GPS on a rat. Instead, it uses rodent complaints as indicators of rat activity and then extrapolates from those reports to generate a probabilistic model of where rats are most likely to be found.
How the Model Works
According to the project's methodology, each NYC 311 rodent complaint is converted into a "probability well" within a city-wide density surface. Subway stations are given additional weight because the underground network is widely regarded as one of the city's most favourable rat habitats, providing shelter, warmth and a reliable food supply.
The model is then adjusted using a circadian activity curve. Since rats are nocturnal, activity levels increase after dark and decline during the day. The site also incorporates a simplified sanitation component based on New York's refuse set-out periods, reflecting the increased availability of food waste during evening and overnight hours.
Understanding the Map
The map itself consists of two distinct visual layers:
The glowing red heat map represents the underlying probability surface generated by the model. Areas shown in bright red indicate locations where the combination of rodent complaints, subway infrastructure and activity factors suggest relatively high rat density. Areas with little or no glow represent places where the model predicts lower rat activity.
The swarm of animated dots represents individual rats swarming around the city in real-time. Obviously these are not real rats. Instead, they are Monte Carlo agents sampled from the underlying density field. In other words, the dots are visual markers drawn from the model's probability distribution.
The heat map represents the model itself; the animated dots are a visual expression of that model.



Comments