Developing Driver Safety Prediction Models

Project Description This dataset was created in support of one of the winners in the "Student Paper Competition: SHRP 2 Safety Data Bonanza" sponsored by the SHRP 2 Safety Data Program and the Transportation Research Board (TRB) Committee for Use and Oversight of SHRP 2 Safety Data, Phase 1. The data enabled the development of driver safety prediction models utilizing driver, vehicle, and trip information. The model data distribution followed the negative binomial distribution. The response variable was crash event severity, and the predictor variables involved behavioral, vehicular, and travel information. The models estimated the expected number of crashes per driver over a period of time. Data Request Scope The dataset consisted of all SHRP 2 crashes, near-crashes, and crash relevant events available within the InSight website at the time of the study, for a total of 3,593 drivers. Trip summary data was for a single year of the study, 2012. Data Specification The dataset included variables from the following InSight tables: Barkley's ADHD Screening Test Data: For variable list, please see BarkleyTestDataDictionary.pdf. Driver Demographic Questionnaire: For variable list, please see DriverDemographicDataDictionary.pdf. Driving History Questionnaire: For variable list, please see DrivingHistoryDataDictionary.pdf. Event Detail Table: For variable list, please see EventTableDataDictionary.pdf. Medical Conditions Medications: For variable list, please see MedicalConditionsDataDictionary.pdf. Sleep Habits Questionnaire: For variable list, please see SleepHabitsDataDictionary.pdf. Trip Summary Table: For variable list, please see TripSummaryDataDictionary.pdf. Vehicle Detail Table: For variable list please see VehicleTableDataDictionary.pdf.