Predicting Freeway Incident Duration Using Machine Learning

Hamad, Khaled; Khalil, Mohamad Ali; Alozi, Abdul Razak

doi:10.1007/s13177-019-00205-1

Cited by 19 publications

(9 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nevertheless, the SVM outperforms all other ML techniques, which was also reported by Hamad et al. (2019).…”

Section: Resultssupporting

confidence: 81%

“…A similar investigation was conducted by Hamad et al. (2019), where the performance of multiple ML approaches including regression trees, SVM, ensemble trees, ANN, and Gaussian process regression (GPR) in predicting incident duration was assessed. The results revealed that SVM and GPR performed the best over other ML techniques in terms of MAE, yet they required the longest training time over other ML techniques.…”

Section: Related Workmentioning

confidence: 98%

See 1 more Smart Citation

Factorial design–machine learning approach for predicting incident durations

Hamad

Obaid

Haridy

et al. 2022

Computer aided Civil Eng

View full text Add to dashboard Cite

This research proposes a hybrid approach for predicting incident duration that integrates the salient features of both factorial design of experiments (DOE) and machine learning (ML). This study compares DOE with another widely used technique, forward sequential feature selection (FSFS). Moreover, to confirm the effectiveness and robustness of the proposed approach, multiple ML techniques are employed, including linear regression, decision trees, support vector machines, ensemble trees, Gaussian process regression, and artificial neural networks. The study results are validated using data from the Houston TranStar incidents archive with over 90,000 records. The accuracy of the developed predictive models is compared based on multiple techniques (i.e., no feature selection–ML, FSFS–ML, and DOE–ML). The results revealed that the significant factors affecting incident duration identified by both DOE and FSFS include the type of vehicles involved, type of lanes affected, number of vehicles involved, number of emergency responses dispatched, incident severity level, and day of the week. The comparative results of the different feature selection and modeling approaches revealed that the hybrid DOE–ML approach outperformed the other tested analysis approaches. The best‐performing model under the DOE–ML approach was the SVM with cubic kernel model. It reduced the modeling time by 83.8% while increasing the prediction error by merely 0.02%, which is not significant. Therefore, the prediction accuracy could be slightly downgraded in return for a substantial reduction in the number of variables utilized, resulting in substantial savings in the modeling time and required dataset.

show abstract

“…Nevertheless, the SVM outperforms all other ML techniques, which was also reported by Hamad et al. (2019).…”

Section: Resultssupporting

confidence: 81%

Section: Related Workmentioning

confidence: 98%

Factorial design–machine learning approach for predicting incident durations

Hamad

Obaid

Haridy

et al. 2022

Computer aided Civil Eng

View full text Add to dashboard Cite

show abstract

“…In recent years, many scholars have reviewed previous machine learning models of event duration [ 29 , 30 ], in which Khaled retrieved more than 110,000 incident records with over 52 variables from the Houston TranStar data archive. Five machine learning algorithms including regression decision tree, SVM, ensemble tree, Gaussian process regression, and artificial neural network (ANN) were compared [ 31 ]. In 2022, Grigorev presented a novel bi-level machine learning framework enhanced with outlier removal and intra–extra joint optimization and found the optimal threshold between short-term versus long-term traffic incident duration.…”

Section: Introductionmentioning

confidence: 99%

Prediction of Duration of Traffic Incidents by Hybrid Deep Learning Based on Multi-Source Incomplete Data

Shang

Xie

2022

IJERPH

View full text Add to dashboard Cite

Traffic accidents causing nonrecurrent congestion and road traffic injuries seriously affect public safety. It is helpful for traffic operation and management to predict the duration of traffic incidents. Most of the previous studies have been in a certain area with a single data source. This paper proposes a hybrid deep learning model based on multi-source incomplete data to predict the duration of countrywide traffic incidents in the U.S. The text data from the natural language description in the model were parsed by the latent Dirichlet allocation (LDA) topic model and input into the bidirectional long short-term memory (Bi-LSTM) and long short-term memory (LSTM) hybrid network together with sensor data for training. Compared with the four benchmark models and three state-of-the-art algorithms, the RMSE and MAE of the proposed method were the lowest. At the same time, the proposed model performed best for durations between 20 and 70 min. Finally, the data acquisition was defined as three phases, and a phased sequential prediction model was proposed under the condition of incomplete data. The results show that the model performance was better with the update of variables.

show abstract

“…A detailed description of the RF model is beyond the scope of this manuscript. Further information on RF methods can be found in Baker and Ellison (2008) and Hamad et al (2019).…”

Section: ) Untangling Vza Sza Amf and Pwvmentioning

confidence: 99%

Advancing Maritime Transparent Cirrus Detection Using the Advanced Baseline Imager “Cirrus” Band

McHardy

Campbell

Peterson

et al. 2021

Journal of Atmospheric and Oceanic Technology

View full text Add to dashboard Cite

We describe a quantitative evaluation of maritime transparent cirrus cloud detection, which is based on Geostationary Operational Environmental Satellite – 16 (GOES-16) and developed with collocated Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) profiling. The detection algorithm is developed using one month of collocated GOES-16 Advanced Baseline Imager (ABI) Channel 4 (1.378 μm) radiance and CALIOP 0.532 μm column-integrated cloud optical depth (COD). First, the relationships between the clear-sky 1.378 μm radiance, viewing/solar geometry, and precipitable water vapor (PWV) are characterized. Using machine learning techniques, it is shown that the total atmospheric pathlength, proxied by airmass factor (AMF), is a suitable replacement for viewing zenith and solar zenith angles alone, and that PWV is not a significant problem over ocean. Detection thresholds are computed using the Ch. 4 radiance as a function of AMF. The algorithm detects nearly 50% of sub-visual cirrus (COD < 0.03), 80% of transparent cirrus (0.03 < COD < 0.3), and 90% of opaque cirrus (COD > 0.3). Using a conservative radiance threshold results in 84% of cloudy pixels being correctly identified and 4% of clear-sky pixels being misidentified as cirrus. A semi-quantitative COD retrieval is developed for GOES ABI based on the observed relationship between CALIOP COD and 1.378 μm radiance. This study lays the groundwork for a more complex, operational GOES transparent cirrus detection algorithm. Future expansion includes an over-land algorithm, a more robust COD retrieval that is suitable for assimilation purposes, and downstream GOES products such as cirrus cloud microphysical property retrieval based on ABI infrared channels.

show abstract

Predicting Freeway Incident Duration Using Machine Learning

Cited by 19 publications

References 31 publications

Factorial design–machine learning approach for predicting incident durations

Factorial design–machine learning approach for predicting incident durations

Prediction of Duration of Traffic Incidents by Hybrid Deep Learning Based on Multi-Source Incomplete Data

Advancing Maritime Transparent Cirrus Detection Using the Advanced Baseline Imager “Cirrus” Band

Contact Info

Product

Resources

About