Crash injury severity analysis using a two-layer Stacking framework.

Crash injury severity analysis is useful for traffic management agency to further understand severity of crashes. A two-layer Stacking framework is proposed in this study to predict the crash injury severity: The fist layer integrates advantages of three base classification methods: RF (Random Forests), AdaBoost (Adaptive Boosting), and GBDT (Gradient Boosting Decision Tree); the second layer completes classification of crash injury severity based on a Logistic Regression model. A total of 5538 crashes were recorded at 326 freeway diverge areas. In the model calibration, several parameters including the number of trees in three base classification methods, learning rate, and regularization coefficient are optimized via a systematic grid search approach. In the model validation, the performance of the Stacking model is compared with several traditional models including the Support Vector Machine (SVM), Multi-Layer Perceptron (MLP) and Random Forests (RF) in the multi classification experiments. The prediction results show that Stacking model achieves superior performance evaluated by two indicators: accuracy and recall. Furthermore, all the factors used in severity prediction are classified into different categories according to their influence on the results, and sensitivity analysis of several significant factors is finally implemented to explore the impact of their value variation on the prediction accuracy.

[1]  David Mahalel,et al.  A NOTE ON ACCIDENT RISK , 1986 .

[2]  K. Kockelman,et al.  Bayesian Multivariate Poisson Regression for Models of Injury Count, by Severity , 2006 .

[3]  Amirfarrokh Iranitalab,et al.  Comparison of four statistical and machine learning methods for crash severity prediction. , 2017, Accident; analysis and prevention.

[4]  Lior Rokach,et al.  Troika - An improved stacking schema for classification tasks , 2009, Inf. Sci..

[5]  F D Bijleveld,et al.  The covariance between the number of accidents and the number of victims in multivariate analysis of accident related outcomes. , 2005, Accident; analysis and prevention.

[6]  Helai Huang,et al.  Support vector machine in crash prediction at the level of traffic analysis zones: Assessing the spatial proximity effects. , 2015, Accident; analysis and prevention.

[7]  Griselda López,et al.  Extracting decision rules from police accident reports through decision trees. , 2013, Accident; analysis and prevention.

[8]  Xin Pei,et al.  A joint-probability approach to crash prediction models. , 2011, Accident; analysis and prevention.

[9]  Sudip Barua,et al.  Multivariate random parameters collision count data models with spatial heterogeneity , 2016 .

[10]  Geoff Holmes,et al.  Accurate Ensembles for Data Streams: Combining Restricted Hoeffding Trees using Stacking , 2010, ACML.

[11]  Cullen Schaffer.  Cross-Validation, Stacking and Bi-Level Stacking: Meta-Methods for Classification Learning , 1994 .

[12]  Karthik C Konduri,et al.  A simultaneous equations model of crash frequency by severity level for freeway sections. , 2013, Accident; analysis and prevention.

[13]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[14]  Fang Liu,et al.  Lane-changes prediction based on adaptive fuzzy neural network , 2018, Expert Syst. Appl..

[15]  Tessa K Anderson,et al.  Kernel density estimation and K-means clustering to profile road accident hotspots. , 2009, Accident; analysis and prevention.

[16]  Yunlong Zhang,et al.  Forecasting of Short-Term Freeway Volume with v-Support Vector Machines , 2007 .

[17]  Mohamed Abdel-Aty,et al.  Using conditional inference forests to identify the factors affecting crash severity on arterial corridors. , 2009, Journal of safety research.

[18]  J. Friedman.  Greedy function approximation: A gradient boosting machine. , 2001 .

[19]  Ramesh Sharda,et al.  Identifying significant predictors of injury severity in traffic accidents using a series of artificial neural networks. , 2006, Accident; analysis and prevention.

[20]  Juan de Oña,et al.  Analysis of traffic accident severity using Decision Rules via Decision Trees , 2013, Expert Syst. Appl..

[21]  Wei Wang,et al.  Using support vector machine models for crash injury severity analysis. , 2012, Accident; analysis and prevention.

[22]  Wei Wang,et al.  Construct support vector machine ensemble to detect traffic incident , 2009, Expert Syst. Appl..

[23]  Ezra Hauer,et al.  Observational Before-After Studies in Road Safety , 1997 .

[24]  Yoav Freund,et al.  Experiments with a New Boosting Algorithm , 1996, ICML.

[25]  Simon Washington,et al.  A simultaneous equations model of crash frequency by collision type for rural intersections , 2009 .

[26]  Zong Tian,et al.  Investigating driver injury severity patterns in rollover crashes using support vector machine models. , 2016, Accident; analysis and prevention.

[27]  Chao Wang,et al.  Predicting accident frequency at their severity levels and its application in site ranking using a two-stage mixed multivariate model. , 2011, Accident; analysis and prevention.

[28]  Ezra Hauer,et al.  Estimation of safety at signalized intersections , 1988 .

[29]  K. El-Basyouny,et al.  A Full Bayesian Multivariate Count Data Model of Collision Severity with Spatial Correlation , 2014 .

[30]  Helai Huang,et al.  A stable and optimized neural network model for crash injury severity prediction. , 2014, Accident; analysis and prevention.

[31]  Mario De Luca,et al.  Using a K-Means Clustering Algorithm to Examine Patterns of Vehicle Crashes in Before-After Analysis , 2013 .

[32]  Keechoo Choi,et al.  A two-stage bivariate logistic-Tobit model for the safety analysis of signalized intersections , 2014 .

[33]  Mohamed Abdel-Aty,et al.  Development of Artificial Neural Network Models to Predict Driver Injury Severity in Traffic Accidents at Signalized Intersections , 2001 .

[34]  Paul Damien,et al.  A multivariate Poisson-lognormal regression model for prediction of crash counts by severity, using Bayesian methods. , 2008, Accident; analysis and prevention.

[35]  Xuedong Yan,et al.  Exploring precrash maneuvers using classification trees and random forests. , 2009, Accident; analysis and prevention.