DataX Team 28 | Topic: Predicting Traffic Safety of Chicago Streets based on the frequency & severity of accidents
Team Members: Semih Bezci (bezsem11@berkeley.edu) Clara Lim (shuqiclara_lim@berkeley.edu) Anna Matsokina (annamatsokina@berkeley.edu) Jonathan Ong (jonathan.ong.6@berkeley.edu) Eric Zhevel (ericzhevel@berkeley.edu)
A study by the World Health Organization reveals that there are approximately 1.25 million annual fatalities from traffic accidents. Despite advancements in traffic safety and road conditions, there is an approximate annual rate of 5.5 million accidents in the United States. Further analysis of accident features such as road conditions, driver behavior and weather can improve our understanding of recent higher crash rates.
This project thus assesses driving risk levels for individual road segments at a given time in Chicago. To achieve this, we first provide a danger rating score for each accident occurrence and then classify streets into various accident risk levels based on aggregated street rating scores of the streets, which considers both the severity and frequency of the accidents. Combining the geospatial data and danger scores of each accident allowed us to generate intuitive visualizations of traffic accidents, and allowed for street safety risk level predictions based on different times of the day.