Traffic Analysis Model with Bayesian Network and Social Media Data: D100 Highway Travel İnformation System

Cihan Çiftçi; Halim Kazan

Makale detayları

Traffic Analysis Model with Bayesian Network and Social Media Data: D100 Highway Travel İnformation System

Yazarlar:

Cihan Çiftçi | Halim Kazan

Yayıncı:

Istanbul University

Ana Başlıklar:

Yöneylem

Dergi:

Journal of Transportation and Logistics Vol.8 Issue 1, pp.48-61

Belge Tipi:

Makale

Yayın Tarihi:

10/2023

Belge Dili:

İngilizce

Document ID (Org No.):

2001298123

pISSN:

2459-1718

Anahtar kelimeler:

Özet:

 The traffic problem in Intelligent Transportation Systems has recently become a very important issue. Thanks to Intelligent Transportation Systems, the formation of large amounts of traffic data has led to the formation of data-oriented models. There is a growing interest in predicting traffic measures by modeling complex scenarios based on big data with data mining and machine learning methods. In this study, traffic events from Twitter traffic notifications and vehicle density from sensor data were obtained. Traffic density analysis and traffic incident analysis were performed with the machine learning method. In the analysis of traffic incidents, 36627 traffic incidents were digitized. This data was separated into categories including type of accident; day; month; year; season; left, right or middle lane; and vehicle failure, maintenance-repair work and accident notification. Between 2016 and 2020, 1400 daily vehicle data logs were obtained from the sensor data located at 59 points of the D100 highway. Traffic density and parameters affecting traffic incidents on the Anatolian and European sides of the D100 highway in Istanbul were determined. Traffic density and accident event models were designed with the Bayesian network approach. In the sensitivity analysis of the model, it was concluded that the parameter that has the strongest effect on traffic events and density formation on the D100 highway line is the strips. With these models, the infrastructure of the early warning system has been created for region-specific traffic density situations and possible traffic events.

 The traffic problem in Intelligent Transportation Systems has recently become a very important issue. Thanks to Intelligent Transportation Systems, the formation of large amounts of traffic data has led to the formation of data-oriented models. There is a growing interest in predicting traffic measures by modeling complex scenarios based on big data with data mining and machine learning methods. In this study, traffic events from Twitter traffic notifications and vehicle density from sensor data were obtained. Traffic density analysis and traffic incident analysis were performed with the machine learning method. In the analysis of traffic incidents, 36627 traffic incidents were digitized. This data was separated into categories including type of accident; day; month; year; season; left, right or middle lane; and vehicle failure, maintenance-repair work and accident notification. Between 2016 and 2020, 1400 daily vehicle data logs were obtained from the sensor data located at 59 points of the D100 highway. Traffic density and parameters affecting traffic incidents on the Anatolian and European sides of the D100 highway in Istanbul were determined. Traffic density and accident event models were designed with the Bayesian network approach. In the sensitivity analysis of the model, it was concluded that the parameter that has the strongest effect on traffic events and density formation on the D100 highway line is the strips. With these models, the infrastructure of the early warning system has been created for region-specific traffic density situations and possible traffic events.

 The traffic problem in Intelligent Transportation Systems has recently become a very important issue. Thanks to Intelligent Transportation Systems, the formation of large amounts of traffic data has led to the formation of data-oriented models. There is a growing interest in predicting traffic measures by modeling complex scenarios based on big data with data mining and machine learning methods. In this study, traffic events from Twitter traffic notifications and vehicle density from sensor data were obtained. Traffic density analysis and traffic incident analysis were performed with the machine learning method. In the analysis of traffic incidents, 36627 traffic incidents were digitized. This data was separated into categories including type of accident; day; month; year; season; left, right or middle lane; and vehicle failure, maintenance-repair work and accident notification. Between 2016 and 2020, 1400 daily vehicle data logs were obtained from the sensor data located at 59 points of the D100 highway. Traffic density and parameters affecting traffic incidents on the Anatolian and European sides of the D100 highway in Istanbul were determined. Traffic density and accident event models were designed with the Bayesian network approach. In the sensitivity analysis of the model, it was concluded that the parameter that has the strongest effect on traffic events and density formation on the D100 highway line is the strips. With these models, the infrastructure of the early warning system has been created for region-specific traffic density situations and possible traffic events.

 The traffic problem in Intelligent Transportation Systems has recently become a very important issue. Thanks to Intelligent Transportation Systems, the formation of large amounts of traffic data has led to the formation of data-oriented models. There is a growing interest in predicting traffic measures by modeling complex scenarios based on big data with data mining and machine learning methods. In this study, traffic events from Twitter traffic notifications and vehicle density from sensor data were obtained. Traffic density analysis and traffic incident analysis were performed with the machine learning method. In the analysis of traffic incidents, 36627 traffic incidents were digitized. This data was separated into categories including type of accident; day; month; year; season; left, right or middle lane; and vehicle failure, maintenance-repair work and accident notification. Between 2016 and 2020, 1400 daily vehicle data logs were obtained from the sensor data located at 59 points of the D100 highway. Traffic density and parameters affecting traffic incidents on the Anatolian and European sides of the D100 highway in Istanbul were determined. Traffic density and accident event models were designed with the Bayesian network approach. In the sensitivity analysis of the model, it was concluded that the parameter that has the strongest effect on traffic events and density formation on the D100 highway line is the strips. With these models, the infrastructure of the early warning system has been created for region-specific traffic density situations and possible traffic events.

Makale detayları

Traffic Analysis Model with Bayesian Network and Social Media Data: D100 Highway Travel İnformation System

Yazarlar:

Cihan Çiftçi | Halim Kazan

Yayıncı:

Istanbul University

Ana Başlıklar:

Yöneylem

Dergi:

Journal of Transportation and Logistics Vol.8 Issue 1, pp.48-61

Belge Tipi:

Makale

Yayın Tarihi:

10/2023

Belge Dili:

İngilizce

Document ID (Org No.):

2001298123

pISSN:

2459-1718

Traffic management

traffic analysis

machine learning

bayesian networks

big data

twitter PDF Görünüm Referanslar Atıflar PDF Görünüm

Sayı

Makale detayları

Traffic Analysis Model with Bayesian Network and Social Media Data: D100 Highway Travel İnformation System

Yazarlar: Cihan Çiftçi | Halim Kazan

Yayıncı: Istanbul University

Ana Başlıklar: Yöneylem

Dergi: Journal of Transportation and Logistics Vol.8 Issue 1, pp.48-61

Belge Tipi: Makale

Yayın Tarihi: 10/2023

Belge Dili: İngilizce

Document ID (Org No.): 2001298123

pISSN: 2459-1718

Traffic management

traffic analysis

machine learning

bayesian networks

big data

twitter PDF Görünüm Referanslar Atıflar PDF Görünüm

Sayı

Yazarlar:

Cihan Çiftçi | Halim Kazan

Yayıncı:

Istanbul University

Ana Başlıklar:

Yöneylem

Dergi:

Journal of Transportation and Logistics Vol.8 Issue 1, pp.48-61

Belge Tipi:

Makale

Yayın Tarihi:

10/2023

Belge Dili:

İngilizce

Document ID (Org No.):

2001298123

pISSN:

2459-1718