In an effort to enhance traffic safety within New York City, our team applied advanced machine learning (ML) techniques to conduct time series forecasting across a wide network of census tracts. 

 

This ML model forecasts the daily number of traffic crashes per tract over a 14-day period. To make these predictions accessible and actionable, we developed an intuitive visualization represented through a color gradient. Darker shades indicate a higher predicted likelihood of experiencing five or more crashes per day, while lighter shades suggest fewer predicted crashes. This gradient not only highlights but also helps prioritize areas with a higher risk, thus guiding targeted interventions.