The City of Philadelphia’s Office of Transportation and Infrastructure Systems’ Office of Multimodal Planning (“the client”) wishes to understand the relationship between roadway safety outcomes and infrastructural, temporal and behavioral factors - specifically in off-peak hours. The goals of this project are to investigate the hypothesis that designing roadways for peak hour vehicle operation is related to adverse safety outcomes during off-peak hours (crashes) and build a model that can estimate the frequency of high speed driving given road characteristics. Put simply - congestion may contribute to fender benders, but excess capacity might be contributing to catastrophic outcomes. Your investigation will be done by modeling the relationships between high-speed driving, injurious crashes, temporal factors, and roadway characteristics. The outcome of interest is subject to determination by the team - since crashes are stochastic in nature, they are difficult to model and forecast as accurately as continuous variables. Frequency of high-speed driving, which is believed to be highly correlated to KSI (killed or serious injury) crashes, is likely to be the outcome of interest for forecasting.
The use case is as follows: Philadelphia transportation planners want a data system that can identify street segments which might benefit from safety interventions designed to reduce the amount of high-speed driving. You will design a predictive model that can estimate the change in the frequency of high-speed driving as a function of changes in road diet, and present this information in an application that can visualize prospective changes.