Fuzzy C-means clustering based on clustering algorithms for traffic crash data [Conference Object]

Abstract

Clustering generally is related to classification problem. This paper used three clustering algorithm-hierarchical, K-means, and fuzzy C-means clustering- to obtain the number of cluster which finally was applied to fuzzy C-means clustering so as to determine the classification accuracy and time response. The dataset used consists of 1049 traffic crashes and was derived from traffic crashes reported between 2005 and 2010 on the North Cyprus primary road network. Seven parameters are selected as input for such a model: driver's gender, driver's age, crash time, type of vehicle, weather condition, traffic way character, and collision type and three injury levels were taken into consideration for this study (i.e. no injury, evident injury, fatality).The results showed that data classification accuracy was around 0.725 and time response was around 0.474 according to fuzzy C-means clustering. © 2013 Taylor & Francis Group, London.