Skip to main content

Visualizing, Modeling, and Forecasting Crashes in Wake County, North Carolina

Presented by:
Ayan Gulzar (North Carolina Central University), Cortyln Morris (Haverford College), Kelly Wentzlof (Indiana University)

Studies have shown that road traffic crashes are one of the leading causes of unnatural death for U.S citizens. In our research, we address three questions about traffic crashes in Wake County, North Carolina: (1) Where are the hotspots for crashes in the Wake County area?; (2) Can we reliably forecast the daily/monthly number of crashes in Wake County? Given the weather and road features and drivers' characteristics, can we predict the crash outcome (whether it results in injuries)? To accomplish this aim, two public datasets covering the period between January 2015 to May 2021 obtained from the North Carolina Wake County Open Data Portal were used (persons involved in crashes and crash locations). The paper answers the above research questions using a combination of exploratory data analysis and visualization techniques, time series forecasting techniques (TBATS, ARIMA, HoltWinters, and a combination of STLF and SARIMA) and predictive modeling techniques (logistic regression and random forest).