Department of Innovation and Technology
City of Chicago, Illinois
Chicago’s Lake Michigan shoreline offers 26 enticing miles of beaches for 20 million users annually, but E. coli outbreaks threaten swimmers with flu-like illnesses. The Chicago Park District uses traditional E.coli lab tests, but they take 18-24 hours to process, too slow to warn swimmers. New, rapid (2-3 hours) DNA tests are effective but expensive. Eradicating E.coli is impossible, but earlier notice is needed to protect the public.
The city worked with a weekly hackathon, Chi Hack Night, and the volunteers created four computer models in spring 2016 to forecast E.coli outbreaks. That summer, the city tried out the models, comparing them with environmental, weather and water data to pinpoint E.coli appearances.
The models proved ineffective themselves, but the city opted for a hybrid approach. It employed machine-learning algorithms (available on GitHub) and same-day DNA testing at key beaches for more accurate results. Coded through R, the open-source programming language, the hybrid effort was three times more accurate in predicting outbreaks and safeguarding swimmers.
Solutions can come in unlikely areas so look for help where you can, even outside conventional avenues. Sometimes the best option is a hybrid that incorporates both new data applications and older approaches.
The city’s lead data scientist running the E.coli initiative for Chicago Park District. He was one of the many volunteer hackers who devised the original models for E.coli forecasting.
A Toledo, Ohio native, planned a life in the law after Ohio State (BA, ‘03) and DePaul (JD, MS, ‘06). Interned in the Ohio House of Representatives, and later worked as a Chicago assistant corporation counsel enforcing building safety codes. His inner computer nerd surfaced, and shifted into data science.
Nick and his brother are living together in an old home with their young families and rehabbing the three-flat in Logan Square for the last three years.