Investigation of time and weather effects on crash types using full Bayesian multivariate Poisson lognormal models

El-Basyouny, Karim; Barua, Sudip; Islam, Tazul

doi:10.1016/j.aap.2014.08.014

Cited by 89 publications

(44 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Apart from developing state-of-the-art approaches for real-time crash risk assessment, researchers have made various attempts to seek for more relevant explanatory variables. El-Basyouny et al (2014) investigated the impact of the sudden extreme snow or rain variation on the crash type, using full Bayesian multivariate Poisson log-normal models. proposed a hierarchical logistic regression model to predict crash likelihood, with multi-source information, including traffic, weather, and roadway geometric factors.…”

Section: Literature Reviewmentioning

confidence: 99%

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

Zhang

Chen

2018

Transportmetrica A: Transport Science

View full text Add to dashboard Cite

The real-time crash likelihood prediction has been an important research topic. Various classifiers, such as support vector machine (SVM) and tree-based boosting algorithms, have been proposed in traffic safety studies. However, few research focuses on the missing data imputation in real-time crash likelihood prediction, although missing values are commonly observed due to breakdown of sensors or external interference. Besides, classifying imbalanced data is also a difficult problem in real-time crash likelihood prediction, since it is hard to distinguish crash-prone cases from non-crash cases which compose the majority of the observed samples. In this paper, principal component analysis (PCA) based approaches, including LS-PCA, PPCA, and VBPCA, are employed for imputing missing values, while two kinds of solutions are developed to solve the problem in imbalanced data. The results show that PPCA and VBPCA not only outperform LS-PCA and other imputation methods (including mean imputation and k-means clustering imputation), in terms of the root mean square error (RMSE), but also help the classifiers achieve better predictive performance. The two solutions, i.e., cost-sensitive learning and synthetic minority oversampling technique (SMOTE), help improve the sensitivity by adjusting the classifiers to * Corresponding author pay more attention to the minority class. Keywords: Real-time crash likelihood prediction, PCA-based missing data imputation, cost-sensitive learning, SMOTE, support vector machine, AdaBoost 1. Introduction Prediction of traffic crash has been a major research topic in transportation safety studies. Crashes, especially on urban expressways, can trigger heavy traffic congestions, impose huge external costs, and reduce the level of service of transportation infrastructures. Therefore, the accurate and reliable prediction of crash risks is critical to the success of proactive safety management strategies on urban expressways. There have been fruitful studies in the domain of the real-time crash likelihood estimation (Abdel-Aty and Pemmanaboina, 2006; Abdel-Aty et al., 2007, 2008; Ahmed and Abdel-Aty, 2012). It has been reported that crash occurrence was affected by four major factors: real-time traffic state, drivers' behavior, environment factors, and road geometry (Ahmed and Abdel-Aty, 2013b). Traditional devices utilized in detecting real-time traffic states are mainly intrusive, e.g., loop detectors. Recently, more non-intrusive traffic detection devices are in use due to their easiness of installation, maintenance, accuracy, and affordable costs. For example, Remote Traffic Microwave Sensors (RTMS) and Automatic Vehicle Identification (AVI) devices provide access to real-time traffic data from multiple sources. In field applications, RTMS simultaneously provide real-time data of flow, time occupancy, and speed.Despite RTMS or other detectors (e.g., AVI devices and loop detectors) have been widely used and successfully applied in traffic operations including the real-time crash likelihood estimat...

show abstract

Section: Literature Reviewmentioning

confidence: 99%

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

Zhang

Chen

2018

Transportmetrica A: Transport Science

View full text Add to dashboard Cite

show abstract

“…In the covariance matrix, the diagonal elements can be larger than the mean to accommodate the over-dispersion. The off-diagonal elements would take the correlation of different crash types of the same record into account (El-Basyouny et al 2014).…”

Section: Methodsmentioning

confidence: 99%

“…Thus, the uninformative priors are used. Referring to the past studies (Ma et al 2008;Park and Lord 2008;El-Basyouny et al 2014), each element of . is assumed to independently follow the (0,1000) distribution, where the variance of 1000 is helpful to find the real distributions of regression coefficients in a big range.…”

Section: Prior Distributionmentioning

confidence: 99%

See 1 more Smart Citation

Multivariate Poisson-Lognormal Model for Analysis of Crashes on Urban Signalized Intersections Approach

Zhao

Liu

et al. 2017

Journal of Transportation Safety & Security

View full text Add to dashboard Cite

Many studies investigate contributing factors of intersection crashes, but very limited studies focus on cashes on the intersection approach. It is important to address the characteristics of intersection-approach crashes to better understand intersection safety. This article analyzes the crashes on signalized intersection approach on urban arterials with a focus on traffic and geometric elements. The intersection approach is defined as the segment between stop bar and the location 200 ft upstream from the stop bar. The multivariate Poisson lognormal (MVPLN) model is used to model crash counts by severity. Ten-year crash data collected from 643 signalized intersections in Nebraska are used for analysis. One-way road is found to be negatively related to all three severity levels (light crash, moderate crash, and severe crash) of crashes. Compared to the 12 ft lane width, narrower lane widths generally lead to fewer crashes. The intersection approaches on urban arterials are expected to have more crashes than collector roads. The numbers of right-turn, left-turn, and through lanes, as well as the annual average daily traffic on the intersection approach and its crossing approach are statistically significant factors increasing crash frequency. The MVPLN model is compared to univariate and zero-inflated Poisson models. The results reveal that the MVPLN model provides a superior fit over the univariate Poisson model. KeywordsMultivariate Poisson log-normal, signalized intersection, crash analysis, traffic safety, intersection approach ABSTRACTMany studies investigate contributing factors of intersection crashes, but few focus on crashes on intersection approaches. It is important to address the characteristics of intersection approach crashes to better understand intersection safety. This paper analyzes the crashes on signalized intersection approaches on urban arterials with a focus on traffic and geometric elements. The intersection approach is defined as the segment between the stop bar and 200 ft upstream, and the multivariate Poisson-lognormal (MVPLN) model is used to model crash counts by severity. Ten-year crash data, collected from 643 signalized intersections in Nebraska, are analyzed. It was found that one-way roads negatively impact all three crash severity levels (light, moderate, and severe), and compared to the 12 ft lane width, narrower lane widths generally lead to fewer crashes. The intersection approaches on urban arterials are expected to have more crashes than collector roads. The amount of right-turn, left-turn, and through lanes, and the annual average daily traffic (AADT) on the intersection approach and its crossing approach are statistically significant factors increasing crash frequency. The MVPLN model is compared to univariate and zero-inflated Poisson models. The results reveal that the MVPLN model provides a superior fit compared to the univariate Poisson model.

show abstract

“…[17]), and also to classify descriptive road accident features, such as driver's behaviour, accident type, or accident severity (e.g. [2,5,16]). …”

Section: Introductionmentioning

confidence: 99%

Spatial and temporal aspects of prior and likelihood data choices for Bayesian models in road traﬃc safety analyses

Nowakowska¹

2016

EiN

View full text Add to dashboard Cite

Article citation info: (*) Tekst artykułu w polskiej wersji językowej dostępny w elektronicznym wydaniu kwartalnika na stronie www.ein.org.pl IntroductionTraffic road safety, as an element of a human-vehicle-road system, has been the subject of scientific and research works for many years. There are many researchers and specialists in a wide range of fields or disciplines who are involved in the process of recognizing and understanding mechanisms related to a road crash. Many theories and models have been elaborated in order to evaluate the level of road traffic threats, as well as to identify circumstances, and cause and effect relationships of road accidents. The research area is extensive and covers: simulation and behavioural research (e.g. [8,9]), elaboration of entropy models (e.g. [1,12]), investigations of road polygons including road surroundings, and traffic and weather conditions (speed in particular) (e.g. [3,10]), as well as exploration and mining of real road accident data (e.g. [15,19]).Statistical methods belong to the most important research techniques utilised in analysing real data. There are two approaches in such an analysis. The first one is a frequentist (also known as classical) approach, in which a random event's probability is assumed to be represented by the frequency of the event occurrence in a very large number of identical samples. The other one is a Bayesian (also known as non-classical) approach, according to which a prior (unconditional) probability of a random event is a measure of a rational belief that the event will occur. Then, the belief is modified using data from experiments or from observations of circumstances connected with the event. Prior knowledge is transformed into posterior NowAkowskA M. spatial and temporal aspects of prior and likelihood data choices for Bayesian models in road traffic safety analyses. Eksploatacja i Niezawodnosc -Maintenance and Reliability 2017; 19 (1): 68-75, http://dx.doi.org/10.17531/ein.2017 knowledge, which is a resultant probability and a measure of a rational expectation of the event occurrence after getting information from the data. Bayesian thinking, supported by the development of numerical sampling techniques, has created modern statistics fundamentals, which enables formulating and solving problems not available in classical statistics. Bayesian regression modelling is a non-classical methodology which becomes widespread in road traffic safety analyses, mainly because it allows eliminating various weaknesses of classical models. Bayesian regression models are difficult from both conceptual and computational points of view. Nevertheless, they bring a new quality to the development of scientific research methods, and they enable a flexible, though non-standard, approach to modelling issues. The models are used in order to develop safety performance functions (e.g. [6,7,13,16]), including a before-after analysis (e.g. [17]), and also to classify descriptive road accident features, such as driver's behaviour, accident type, or accid...

show abstract

Investigation of time and weather effects on crash types using full Bayesian multivariate Poisson lognormal models

Cited by 89 publications

References 35 publications

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

Multivariate Poisson-Lognormal Model for Analysis of Crashes on Urban Signalized Intersections Approach

Spatial and temporal aspects of prior and likelihood data choices for Bayesian models in road traﬃc safety analyses

Contact Info

Product

Resources

About