Real-time traffic crash prediction is very important to traffic areas because it can help take active intervention actions before traffic accidents happen, and hence reduce accident rates. This study aims to use ensemble learning methods to improve the real-time traffic crash prediction accuracy by applying them to build models based on traffic flow data with crash occurrence labels collected from the Shanghai urban expressway in April and May of 2014. We applied four ensemble methods to build models, and used AUC, F-measure, and other evaluating indicators to measure the classification result of these models. It was concluded that the best ensemble learning model based on decision tree and logistic regression are random forest and LogitBoost, respectively. Their AUC measure can reach to 0.703 and 0.733, which are 7.99% and 4.12% higher than individual learners.
Utilizing Ensemble Learning Methods in Real-Time Traffic Crash Prediction
18th COTA International Conference of Transportation Professionals ; 2018 ; Beijing, China
CICTP 2018 ; 2179-2185
2018-07-02
Conference paper
Electronic Resource
English
Real-Time Crash Prediction Model for Application to Crash Prevention in Freeway Traffic
Transportation Research Record | 2003
|Real-Time Crash Prediction Model for Application to Crash Prevention in Freeway Traffic
Online Contents | 2003
|Comparing Machine Learning and Deep Learning Methods for Real-Time Crash Prediction
Transportation Research Record | 2019
|TSDCN: Traffic safety state deep clustering network for real‐time traffic crash‐prediction
DOAJ | 2021
|