Title: An analysis of the most accident prone regions within the Dhaka Metropolitan Region using clustering

Authors: Shuvashish Paul; Ashik Mostafa Alvi; Rashedur M. Rahman

Addresses: Department of Electrical and Computer Engineering, North South University, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Bashundhara, Dhaka, Bangladesh ' Department of Electrical and Computer Engineering, North South University, Bashundhara, Dhaka, Bangladesh

Abstract: Most of the world's developed countries have decreased the unusual deaths like traffic accidents of their citizens by taking efficient steps. In Bangladesh, injuries because of road accidents have become a regular incident. The highly-populated cities in Bangladesh are still having such incidents daily. As the number of vehicles is increasing and most of the drivers are not willing to follow the traffic rules, injuries due to traffic accidents are not going down at all. Among all those big cities in Bangladesh, Dhaka City has the highest amount of road accidents. So, in this paper we focus on the most hazardous regions in Dhaka Metropolitan area. We have collected the accident related data from Accident Research Institute (ARI) at Bangladesh University of Engineering and Technology (BUET) that is located in the city of Dhaka. In our paper, we have used the fuzzy C-means clustering, expectation maximisation, hierarchical agglomerative clustering and K-means clustering to identify the regions where traffic incidents occurs the most in Dhaka Metropolitan area. The missing values for some attributes in the dataset are overwritten by the mean/mode of that attribute itself.

Keywords: data mining; accidental injury severity; clustering; hazardous areas; Dhaka Metropolitan area.

DOI: 10.1504/IJAIP.2021.113324

International Journal of Advanced Intelligence Paradigms, 2021 Vol.18 No.3, pp.294 - 315

Received: 15 Jun 2016
Accepted: 15 Feb 2017
Published online: 01 Mar 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article