Modeling Traffic Incident Duration Using Quantile Regression

Khattak, Asad J.; Li, Jun; Wali, Behram; Li, Xiaobing; Ng, ManWo

doi:10.3141/2554-15

Cited by 60 publications

(31 citation statements)

References 35 publications

(29 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The existence of multicollinearity among independent variables was checked by VIFs (Table 2). The VIF values of each variable are much smaller than 10, indicating the absence of significant multicollinearity [41].…”

Section: Resultsmentioning

confidence: 99%

Injury Severity and Contributing Driver Actions in Passenger Vehicle–Truck Collisions

Wali

et al. 2019

IJERPH

Self Cite

View full text Add to dashboard Cite

Large-scale truck-involved crashes attract great attention due to their increasingly severe injuries. The majority of those crashes are passenger vehicle–truck collisions. This study intends to investigate the critical relationship between truck/passenger vehicle driver’s intentional or unintentional actions and the associated injury severity in passenger vehicle–truck crashes. A random-parameter model was developed to estimate the complicated associations between the risk factors and injury severity by using a comprehensive Virginia crash dataset. The model explored the unobserved heterogeneity while controlling for the driver, vehicle, and roadway factors. Compared with truck passengers, occupants in passenger vehicles are six times and ten times more likely to suffer minor injuries and serious/fatal injuries, respectively. Importantly, regardless of whether passenger vehicle drivers undertook intentional or unintentional actions, the crashes are more likely to associate with more severe injury outcomes. In addition, crashes occurring late at night and in early mornings are often correlated with more severe injuries. Such associations between explanatory factors and injury severity are found to vary across the passenger vehicle–truck crashes, and such significant variations of estimated parameters further confirmed the validity of applying the random-parameter model. More implications based on the results and suggestions in terms of safe driving are discussed.

show abstract

Section: Resultsmentioning

confidence: 99%

Injury Severity and Contributing Driver Actions in Passenger Vehicle–Truck Collisions

Wali

et al. 2019

IJERPH

Self Cite

View full text Add to dashboard Cite

show abstract

“…From the perspective of incident duration modeling, a broad spectrum of studies has focused on analyzing traffic incidentsspecifically incident durations-to identify key factors associated with incidents for better incident management strategies (5)(6)(7)(8). From a methodological standpoint, incident durations and associated factors have been modeled successfully using a diverse set of rigorous statistical tools such as truncated and quantile regression (9,10), hazard-based duration models (6,11), Bayesian network tools (12)(13)(14), artificial neural networks (15,16), text analysis and competing risk models (17,18), and recently finite mixture models (19), among others. Several correlates such as accident and injury involvement, lane closure, number of vehicles, temporal and spatial factors, heavy-truck involvement, and adverse weather were found positively associated with longer incident durations (6,10,13,14).…”

mentioning

confidence: 99%

“…From a methodological standpoint, incident durations and associated factors have been modeled successfully using a diverse set of rigorous statistical tools such as truncated and quantile regression (9,10), hazard-based duration models (6,11), Bayesian network tools (12)(13)(14), artificial neural networks (15,16), text analysis and competing risk models (17,18), and recently finite mixture models (19), among others. Several correlates such as accident and injury involvement, lane closure, number of vehicles, temporal and spatial factors, heavy-truck involvement, and adverse weather were found positively associated with longer incident durations (6,10,13,14). A paper by Zhang et al contains a summary of findings from different studies (3).…”

mentioning

confidence: 99%

Role of Multiagency Response and On-Scene Times in Large-Scale Traffic Incidents

Khattak

Wali

2017

Transportation Research Record: Journal of the Transportation R

Self Cite

View full text Add to dashboard Cite

Traffic incidents, often known as nonrecurring events, impose enormous economic and social costs. Compared with short-duration incidents, large-scale incidents can substantially disrupt traffic flow by blocking lanes on highways for long periods. A careful examination of large-scale traffic incidents and associated factors can assist with actionable largescale incident management strategies. For such an analysis, a unique and comprehensive 5-year incident database on East Tennessee roadways was assembled to conduct an in-depth investigation of large-scale incidents, especially focusing on operational responses, that is, response and on-scene times by various agencies. Incidents longer than 120 min andblocking at least one lane were considered large scale; the database contained 890 incidents, which was about 0.69% of all reported incidents. Rigorous fixed-and random-parameter, hazard-based duration models were estimated to account for the possibility of unobserved heterogeneity in large-scale incidents. The modeling results reveal significant heterogeneity in associations between operational responses and large-scale incident durations. A 30-min increase in response time for the first, second, and third (or more) highway response units translated to a 2.8%, 1.6%, and 4.2% increase in large-scale incident durations, respectively. In addition, longer response times for towing and highway patrol were significantly associated with longer incident durations. Given large-scale incidents, associated factors included vehicle fire, unscheduled roadwork, weekdays, afternoon peaks, and traffic volume. Notably, the associations were heterogeneous; that is, the direction could be positive in some cases and negative in others. Practical implications of the results for large-scale incident management are discussed.

show abstract

“…A wide range of studies exist regarding traffic incident duration, and various advanced models have recently been applied to increase incident duration prediction accuracy. For example, Khattak et al (2016) used a quantile regression model by capturing variations between different quantiles of incident duration (11), instead of modeling based off mean of duration. Another study proposed to combine survival models with text analysis to obtain high duration prediction accuracy (12), namely because some critical incident-related information is embedded in the narrative text (e.g., involving agencies).…”

Section: Literature Reviewmentioning

confidence: 99%

Sequential Prediction for Large-Scale Traffic Incident Duration: Application and Comparison of Survival Models

Khattak

et al. 2020

Transportation Research Record

Self Cite

View full text Add to dashboard Cite

A quick and accurate traffic incident duration prediction could greatly facilitate traffic incident management. However, at the very early stage of an incident, limited information is available for prediction. Information gathering for large-scale traffic incidents is a chronological process when a multi-agency response is required. At the early stage, information such as incident start time and roadway and weather conditions may be available, but information about response agencies and incident management solutions (e.g., lane closures) remains unknown. The objective of this study is to develop a sequential prediction method to handle the chronological process of incident information gathering. The method is based upon parametric survival modeling, which is often utilized to predict incident duration. This study took advantage of a unique incident database and identified over 600 large-scale incidents in the East Tennessee area from 2015 to 2016. A five-stage prediction method is proposed according to the chronological process by which information becomes available during incident operations. Using the data, this study compared three survival models: frailty model, multilevel mixed-effects model, and finite mixture model. Generally, with more information becoming available for modeling from the first to the last stage, the models’ performance improved according to the root mean square error and mean absolute percent error. The finite mixture model outperforms the other two models and its mean absolute percentage error is between 10% and 15%. Incident-associated factors at each stage are discussed and implications based on the study outcomes are also covered in the paper.

show abstract

Modeling Traffic Incident Duration Using Quantile Regression

Cited by 60 publications

References 35 publications

Injury Severity and Contributing Driver Actions in Passenger Vehicle–Truck Collisions

Injury Severity and Contributing Driver Actions in Passenger Vehicle–Truck Collisions

Role of Multiagency Response and On-Scene Times in Large-Scale Traffic Incidents

Sequential Prediction for Large-Scale Traffic Incident Duration: Application and Comparison of Survival Models

Contact Info

Product

Resources

About