Open AccessPSO-ANN in preventing traffic collisions: a comparative study
Author(s) -
Md. Ashikuzzaman,
Wasim Akram,
Md. Mydul Islam Anik,
Mahamudul Hasan,
Md Sawkat Ali,
Taskeed Jabid
Publication year - 2021
Publication title -
indonesian journal of electrical engineering and computer science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.241
H-Index - 17
eISSN - 2502-4760
pISSN - 2502-4752
DOI - 10.11591/ijeecs.v24.i3.pp1796-1803
Subject(s) - particle swarm optimization , artificial neural network , centroid , computer science , naive bayes classifier , traffic accident , transport engineering , collision , traffic congestion , intelligent transportation system , artificial intelligence , engineering , machine learning , computer security , support vector machine
Traffic accident is a global threat which causes health and economic casualties all around the world. Due to the expansion of transportation systems, congestion can lead to spike road accident. Every year thousands of people have died due to traffic accidents. Various technologies have been adopted by modern cities to minimize traffic accidents. Therefore, to ensure people’s safety, the concept of the smart city has been introduced. In a smart city, factors like road, light, and weather conditions are important to consider to predict traffic mishap. Several machine learning models have been implemented in the existing literature to determine and predict traffic collision. But the accuracy is not enough and there exist a lot of challenges in determining the accident. In this paper, an approach of particle swarm optimization with artificial neural network (PSO-ANN) has been proposed to determine traffic collision using the dataset of the transport department of United Kingdom. The performance of PSO-ANN outperforms the existing machine learning model. PSO-ANN model can be adopted in the transportation system to counter traffic accident issues. Random Forest, Naıve Bayes, Nearest Centroid, K-Nearest Neighbor classification have been used to compare with the proposed PSO-ANN model.