@article{ref1,
title="Traffic event analysis modeling with machine learning methods using social media data",
journal="International journal for traffic and transport engineering (Belgrade)",
year="2022",
author="Çiftçi, Cihan and Kazan, Halim",
volume="12",
number="4",
pages="464-479",
abstract="There is increasing interest in predicting traffic measures by modeling big data-driven complex scenarios with data mining and machine learning methods. In this study, the parameters of the traffic analysis model were created using 35,697 Twitter traffic notifications. The relationships and effects between the parameters of hour, day, month, season, year, lane, accident status, traffic events were revealed. By using the chi-square method, significant relationships were obtained between the day, time, month, season, and lane parameters of the traffic incidents on the D100 highway. Traffic events analysis machine learning methods of the D100 highway, which has a very important place in Istanbul traffic, were carried out. The traffic events prediction accuracy values of the created model were obtained as Naive Bayes 91.2%, Bayes 91.0% and Artificial Neural Network 93.1%. It has been concluded that accident events occur mostly on Fridays, vehicle breakdowns and maintenance-repair works occur mostly on Thursdays, accident events, vehicle breakdowns and maintenance-repair works mostly occur in the right lane. It was concluded that noon times as time, Thursday and Friday as days, January and July as months, winter season as season and right lane as lane are important parameters in terms of ensuring D100 road traffic safety.<p /> <p>Language: en</p>",
language="en",
issn="2217-5652",
doi="10.7708/ijtte2022.12(4).03",
url="http://dx.doi.org/10.7708/ijtte2022.12(4).03"
}