Congestion caused by traffic crashes can cause serious problems for the road, traffic participants, and the environment. Truck crashes usually have the characteristics of high severity and long duration. Here, we study 12,259 truck crashes in Shanxi Province from 2012 to 2017. An XGBoost algorithm was used to predict crash duration using data after classification and feature selection. Samples were divided into two parts in each group, 80% of samples were used to train model and 20% of samples were used to test model. Root-mean-square error (RMSE) representing the variance between predictive value and true value was used to evaluate models.The model performed well for for truck crashes lasting less than 360 min, with RMSE of 10.8932 for 0–90 min duration and RMSE of 17.6550 for 90–360 min A large RMSE was measured when predicting truck crashes lasting more than 360 min.
Duration Prediction for Truck Crashes Based on the XGBoost Algorithm
19th COTA International Conference of Transportation Professionals ; 2019 ; Nanjing, China
CICTP 2019 ; 5021-5031
2019-07-02
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
PV Power Prediction Based on XGBoost Algorithm
IEEE | 2023
|NTIS | 1998
|Truck Mirrors, Fields of View, and Serious Truck Crashes
NTIS | 2007
|Analysis of Fatal Large Truck Crashes
NTIS | 2003
|