Analysis of Gap Acceptance Behavior for Unprotected Right and Left Turning Maneuvers at Signalized Intersections using Data Mining Methods: A Driving Simulation Approach

Abstract

Gap acceptance predictability has become a vital area of interest for traffic safety and operations due to its complexity and significance in understanding a population’s driving behavior. Recent studies have implemented statistical modeling techniques, such as binary logit model (BLM), to predict drivers’ gap acceptance behaviors. However, these models have inherent presumptions and pre-set correlations that, if contravened, can produce erroneous results. The use of non-parametric data mining techniques, such as decision trees, avoids these deficiencies, thus resulting in improvements to the predictive capability of the models. In this study, the feasibility of C4.5 decision trees, instance-based (IB), and random forest (RF) models for predicting drivers’ gap decisions was examined by comparing their results with BLM. To accomplish this objective, 66 study participants drove through ten driving simulation scenarios requiring the navigation of unprotected right and left-turning maneuvers at four-legged, signalized intersections. The data collected from these tests will provide means to directly compare and rank the data mining and statistical models, while also allowing for the identification of variables that are significantly influencing gap acceptance. Results produced from the models indicated that data mining models were superior to BLM at accurately predicting a participant’s gap decisions. RF models outperformed the C4.5 and IB models in predicting gap acceptance behaviors for both the left and right turning scenarios. Because of its superior performance, the authors recommend the implementation of the RF model for predicting gap decisions at unprotected signalized intersections.

Implementing Artificial Neural Network-Based Gap Acceptance Models in the Simulation Model of a Traffic Circle in SUMO

“…On the other hand, the probabilistic method takes into account the inconsistency and heterogeneity in the gap acceptance behavior of drivers in the minor stream and treats the critical gap as a random variable. There are different methods of calculating the critical gap in the literature, which includes Raff’s method, the clearing time method, the binary probit model, the maximum likelihood method, and so forth, as well as machine learning methods, such as the decision tree (DT), random forest (RF), support vector machine (SVM), and ANN methods ( 13 – 27 ). For example, Sangole and Patil ( 21 ) applied an adaptive neuro-fuzzy interface system (ANFIS) to model drivers’ gap acceptance behavior at a limited priority T-intersection.…”

“…Their results indicated that both the SVM and RF techniques produced roughly similar results, while the RF model performed better than the two others with respect to prediction accuracy and computational costs. In addition, the machine learning techniques proved to be more advantageous than the statistical techniques because, firstly, they can resolve the outliers and missing values problem and, secondly, they are non-parametric tools that do not need any predefined correlation between independent and target variables ( 16 ). Pawar et al ( 17 ) compared the SVM model and binary logit model (BLM) to predict the critical gaps.…”

Systemic Opportunities to Improve Older Pedestrian Safety: Merging Crash Data Analysis and a Stakeholder Workshop

“…To determine variable importance in regard to older pedestrian fatal and serious injury crashes, a random forest analysis was conducted. The use of a random forest, or other machine learning method, to identify important predictors and/or complement traditional models has become prevalent in transportation safety literature (17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). The current study applied this approach to identify important variables in predicting older pedestrian serious injury crashes.…”

Improved Support Vector Machine Models for Work Zone Crash Injury Severity Prediction and Analysis

“…Compared with an ordered probit model with a prediction accuracy of 59.5%, results from the ANN produced a more accurate prediction rate with 74.6% accuracy. In another study, the prediction performance of random forest and decision tree were compared with a binary logit model for predicting drivers’ gap decisions ( 20 ). The results illustrated that nonparametric data mining models were superior to the binary logit model at prediction accuracy.…”