In collaboration with DataKind and Seattle's Vision Zero traffic safety program,
We have done exploratory data analyses and predictive modeling of traffic intensity and crash risk for accidents involving pedestrians and cyclists, using historical collision and vehicle count data as well as census data and road network characteristics. The modeling was done both for mid-block crashes, as well as collisions at intersections. Resulting regression models provide insight into the types of locations and combinations of factors that contribute to the occurrence of crashes and to their severity.
|