A multivariate Poisson-lognormal regression model for prediction of crash counts by severity, using Bayesian methods (original) (raw)
Related papers
Multivariate Poisson-Lognormal Models for Jointly Modeling Crash Frequency by Severity
Transportation Research Record: Journal of the Transportation Research Board, 2007
A new multivariate approach is introduced for jointly modeling data on crash counts by severity on the basis of multivariate Poisson-lognormal models. Although the data on crash frequency by severity are multivariate in nature, they have often been analyzed by modeling each severity level separately, without taking into account correlations that exist among different severity levels. The new multivariate Poisson-lognormal regression approach can cope with both overdispersion and a fully general correlation structure in the data, as opposed to the recently suggested multivariate Poisson regression approach, which allows for neither overdispersion nor a general correlation structure in the data. The new method is applied to the multivariate crash counts obtained from intersections in California for 10 years. The results show promise toward the goal of obtaining more accurate estimates by accounting for correlations in the multivariate crash counts and overdispersion.
Accident; analysis and prevention, 2018
The safety and operational improvements of corridors have been the focus of many studies since they carry most traffic on the road network. Estimating a crash prediction model for total crash counts identifies the crash risk factors that are associated with crash counts at a specific type of road entity. However, this may not reveal useful information to detect the road problems and implement effective countermeasures. Therefore, investigating the contributing factors for crash counts by different types is of great importance. This study aims to provide a good understanding of the contributing factors to crash counts by different types at intersections and roadway segments along corridors. Data from 255 signalized intersections and 220 roadway segments along 20 corridors have been used for this study. The investigated crash types include same direction, angle and turning, opposite direction, non-motorized, single vehicle, and other multi-vehicle crashes. Two models have been estimat...
A simultaneous equations model of crash frequency by severity level for freeway sections
Accident Analysis & Prevention, 2013
This paper presents a simultaneous equations model of crash frequencies by severity level for freeway sections using five-year crash severity frequency data for 275 multilane freeway segments in the State of Washington. Crash severity is a subject of much interest in the context of freeway safety due to higher speeds of travel on freeways and the desire of transportation professionals to implement measures that could potentially reduce crash severity on such facilities. This paper presents a joint Poisson regression model with multivariate normal heterogeneities using the method of Maximum Simulated Likelihood Estimation (MSLE). MSLE serves as a computationally viable alternative to the Bayesian approach that has been adopted in the literature for estimating multivariate simultaneous equations models of crash frequencies. The empirical results presented in this paper suggest the presence of statistically significant error correlations across crash frequencies by severity level. The significant error correlations point to the presence of common unobserved factors related to driver behavior and roadway/traffic/environmental characteristics that influence crash frequencies of different severity levels. In addition, the empirical results show that observed factors generally do not have the same impact on crash frequencies at different levels of severity.
Analytic Methods in Accident Research, 2019
Recent literature on highway safety research has focused on methodological advances to minimize misspecifications and the potential for erroneous estimates and invalid statistical inferences. To further these efforts, this study carries out an empirical assessment of uncorrelated and correlated random-parameters count models for analyzing road crash frequencies on multilane highways considering two crash severities; injury and no-injury. The empirical results indicate that the relative statistical performance of these models is comparable; however, the correlated randomparameters approach accounts for both the heterogeneous effects of explanatory factors across the road segments and the cross-correlations among the random parameter estimates. As noted in the results, statistically significant correlation effects among the random parameters confirm the adequacy of this approach. The safety models for multilane roadways presented in this study can be useful in (i) the detection of critical risk factors on these road types, (ii) the assessment of crash reduction due to improvements in pavement condition and retrofitting of roadway geometric features and, (iii) the prediction of crash frequency while comparing different design alternatives. As such, the outcomes of this study may assist design engineers and highway agencies in designing new or calibrating existing multilane roadways from a safety standpoint.
On the nature of over-dispersion in motor vehicle crash prediction models
Accident Analysis and Prevention, 2007
Statistical modeling of traffic crashes has been of interest to researchers for decades. Over the most recent decade many crash models have accounted for extra-variation in crash counts-variation over and above that accounted for by the Poisson density. The extra-variation -or dispersion -is theorized to capture unaccounted for variation in crashes across sites. The majority of studies have assumed fixed dispersion parameters in over-dispersed crash models-tantamount to assuming that unaccounted for variation is proportional to the expected crash count. . Modeling traffic crash-flow relationships for intersections: dispersion parameter, functional form, and Bayes versus empirical Bayes methods. Transport. Res. Rec. 1840, 31-40] challenged the fixed dispersion parameter assumption, and examined various dispersion parameter relationships when modeling urban signalized intersection accidents in Toronto. They suggested that further work is needed to determine the appropriateness of the findings for rural as well as other intersection types, to corroborate their findings, and to explore alternative dispersion functions.
Accident Analysis & Prevention, 2017
This study aims at contributing to the literature on pedestrian and bicyclist safety by building on the conventional count regression models to explore exogenous factors affecting pedestrian and bicyclist crashes at the macroscopic level. In the traditional count models, effects of exogenous factors on non-motorist crashes were investigated directly. However, the vulnerable road users' crashes are collisions between vehicles and nonmotorists. Thus, the exogenous factors can affect the non-motorist crashes through the non-motorists and vehicle drivers. To accommodate for the potentially different impact of exogenous factors we convert the non-motorist crash counts as the product of total crash counts and proportion of non-motorist crashes and formulate a joint model of the negative binomial (NB) model and the logit model to deal with the two parts, respectively. The formulated joint model is estimated using non-motorist crash data based on the Traffic Analysis Districts (TADs) in Florida. Meanwhile, the traditional NB model is also estimated and compared with the joint model. The result indicates that the joint model provides better data fit and can identify more significant variables. Subsequently, a novel joint screening method is suggested based on the proposed model to identify hot zones for non-motorist crashes. The hot zones of non-motorist crashes are identified and divided into three types: hot zones with more dangerous driving environment only, hot zones with more hazardous walking and cycling conditions only, and hot zones with both. It is expected that the joint model and screening method can help decision makers, transportation officials, and community planners to make more efficient treatments to proactively improve pedestrian and bicyclist safety.
Modeling Intersection Crash Counts and Traffic Volume
1997
This research explored the feasibility of modeling crash counts at intersections with use of available exposure measures. The basic purpose of "exposure" is to serve as a size factor to allow comparison of crash counts among populations of different sizes. In the context of highway crash studies, at first glance, vehicle miles of travel (VMT) appears to be a natural exposure measure. However, VMT is closely related to traffic density and this raises doubts if it can serve the intended purpose of an exposure measure. Data from four-leg signalized intersections in Washtenaw County, Michigan, and the states of California and Minnesota were used in this study. Traffic volumes on the approaches are the routinely available exposure measure. It was noted that in these data sets the same values of traffic volume were often "carried over" several intersections. Using such values of traffic volumes as measures of exposure results in correlations between errors of the indep...
IATSS Research, 2012
The purpose of the study was to compare the prediction power of a simplified non-canonical Poisson crashprediction model to other model types. The model, fitted to serious and fatal crash data from 86 two-lane low-volume rural highway segments, showed a good fit, which was not significantly different from that of a negative binomial model. The application of the present model uses the linear form of the non-canonical Poisson model. Hence the simplification of the model versus other models results from the finding that the expected number of crashes per 1 km is directly proportional to the daily volume, unlike logarithmic functions in other models. In the non-canonical model, it is necessary to estimate only one parameter, whereas estimations of more parameters are needed in the negative binomial model.
Accident prediction models with random corridor parameters
Accident Analysis & Prevention, 2009
Recent research advocates the use of count models with random parameters as an alternative method for analyzing accident frequencies. In this paper a dataset composed of urban arterials in Vancouver, British Columbia, is considered where the 392 segments were clustered into 58 corridors. The main objective is to assess the corridor effects with alternate specifications. The proposed models were estimated in a Full Bayes context via Markov Chain Monte Carlo (MCMC) simulation and were compared in terms of their goodness of fit and inference. A variety of covariates were found to significantly influence accident frequencies. However, these covariates resulted in random parameters and thereby their effects on accident frequency were found to vary significantly across corridors. Further, a Poisson-lognormal (PLN) model with random parameters for each corridor provided the best fit. Apart from the improvement in goodness of fit, such an approach is useful in gaining new insights into how accident frequencies are influenced by the covariates, and in accounting for heterogeneity due to unobserved road geometrics, traffic characteristics, environmental factors and driver behavior. The inclusion of corridor effects in the mean function could also explain enough variation that some of the model covariates would be rendered non-significant and thereby affecting model inference.
Accident; analysis and prevention, 2018
According to crash configuration and pre-crash conditions, traffic crashes are classified into different collision types. Based on the literature, multi-vehicle crashes, such as head-on, rear-end, and angle crashes, are more frequent than single-vehicle crashes, and most often result in serious consequences. From a methodological point of view, the majority of prior studies focused on multivehicle collisions have employed univariate count models to estimate crash counts separately by collision type. However, univariate models fail to account for correlations which may exist between different collision types. Among others, multivariate Poisson lognormal (MVPLN) model with spatial correlation is a promising multivariate specification because it not only allows for unobserved heterogeneity (extra-Poisson variation) and dependencies between collision types, but also spatial correlation between adjacent sites. However, the MVPLN spatial model has rarely been applied in previous research ...