Using Design of Experiment Methods and Multiple Linear Regression to Develop a System-Wide Traffic Fatality and Severe-Injury Prediction Model
Open AccessAbstract of PraxisUsing Design of Experiment Methods and Multiple Linear Regression to Develop a System-Wide Traffic Fatality and Severe-Injury Prediction ModelMillions of crashes involving a severe injury or fatality occur in traffic systems each year. Despite this staggering number, there are limited tools available to predict annual severe crash totals and few that ascertain contributions of system-wide factors independently and collectively. Current models focus attention on specific roadway segments, conduct residual analysis a priori to estimate the reduced number of total crashes, or use macroeconomic indicators and crash trends to forecast future system-wide fatalities. But the models generally fail to consider a systems-level approach that accurately accounts for component interactions to estimate future severe crashes. This research fills the gap by applying well-known systems engineering tools (multiple linear regression and design of experiments methods), by leveraging historical crash data, and by adapting a degradation (or compliance) framework to develop a model for predicting crashes involving a severe injury or fatality. The proposed Degradation Impact Model is demonstrated using California traffic data. It is shown to predict as well as other models found in the literature and provides insights into the role of component degradation on crash severity. Furthermore, the Degradation Impact Model can be updated as a system evolves or changes over time, expanded to decompose component factors and tease out more specific system failures, and adapted for other applications in which operational compliance among components is essential.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.