Evaluation of Parametric and Nonparametric Statistical Models in Wrong-way Driving Crash Severity Prediction
Doctor of Philosophy (PhD)
First Advisor's Name
First Advisor's Committee Title
Second Advisor's Name
Second Advisor's Committee Title
Third Advisor's Name
Third Advisor's Committee Title
Fourth Advisor's Name
Fourth Advisor's Committee Title
Fifth Advisor's Name
B. M. Golam Kibria
Fifth Advisor's Committee Title
Classification and Regression Trees (CART); Clustering; Data Mining; Decision Rules; Decision Tree; Machine Learning; Random Forests; Wrong-way Driving; Traffic; Transportation Engineering; Parametric; Nonparametric; Statistical Model
Date of Defense
Wrong-way driving (WWD) crashes result in more fatalities per crash, involve more vehicles, and cause extended road closures compared to other types of crashes. Although crashes involving wrong-way drivers are relatively few, they often lead to fatalities and serious injuries. Researchers have been using parametric statistical models to identify factors that affect WWD crash severity. However, these parametric models are generally based on several assumptions, and the results could generate numerous errors and become questionable when these assumptions are violated. On the other hand, nonparametric methods such as data mining or machine learning techniques do not use a predetermined functional form, can address the correlation problem among independent variables, display results graphically, and simplify the potential complex relationship between the variables.
The main objective of this research was to demonstrate the applicability of nonparametric statistical models in successfully identifying factors affecting traffic crash severity. To achieve this goal, the performance of parametric and nonparametric statistical models in WWD crash severity prediction was evaluated. The following parametric methods were evaluated: Logistic Regression (LR), Ridge Regression (RR), Least Absolute Shrinkage and Selection Operator (LASSO), Linear Discriminant Analysis (LDA), and Gaussian Naïve Bayes (GNB). The following nonparametric methods were evaluated: Random Forests (RF), Decision Trees (DT), and Support Vector Machine (SVM). The evaluation was based on sensitivity, specificity, and prediction accuracy. The research also demonstrated the applicability of nonparametric supervised learning algorithms on crash severity analysis by combining tree-based data mining techniques and marginal effect analysis to show the correlation between the response and the predictor variables.
The analysis was based on 1,475 WWD crashes that occurred on arterial road networks from 2012-2016 in Florida. The results showed that nonparametric models provided better prediction accuracy on predicting serious injury compared to parametric models. By conducting prediction accuracy comparison, contributor variables’ marginal effect analysis, variable importance evaluation, and crash severity pattern recognition analysis, the nonparametric models have been demonstrated to be valid and proved to serve as an alternative tool in transportation safety studies.
The results showed that head-on collisions, weekends, high-speed facilities, crashes involving vehicles entering from a driveway, dark-not lighted roadways, older drivers, and driver impairment are important factors that play a crucial role in WWD crash severity on non-limited access facilities. This information may assist researchers and safety engineers in identifying specific strategies to reduce the severity of WWD crashes on arterial streets. Besides unveiling the factors contributing to WWD crash severity and their relationship with each other, this research has demonstrated the potential of using data mining techniques in yielding results that are easily understandable and interpretable.
Nafis, Sajidur Rahman, "Evaluation of Parametric and Nonparametric Statistical Models in Wrong-way Driving Crash Severity Prediction" (2021). FIU Electronic Theses and Dissertations. 4620.
Civil Engineering Commons, Computational Engineering Commons, Computer Sciences Commons, Data Science Commons, Other Applied Mathematics Commons, Statistics and Probability Commons, Transportation Engineering Commons
In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/
This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).