Traffic Accident Data Profiling and Clusteringwith Data Mining Process

Abstract

Some of this traffic accident crisis in Nigeria is caused by the increase in number of vehicles and inefficient drivers on the road, bad condition and poor maintenance of the roads. The significance of the study lies on the profiling of clusters of traffic roads in terms of accident related data and the degree in which these accident characteristics are perceptive between the different created clusters. Applying data mining process to model traffic accident data records helped in obtaining the characteristics of drivers' behaviour, road condition and weather condition that are connected with different injury severities and death. The traffic roads are divided into a low accident risk and high accident risk traffic roads, determining accidents in different age categories and period of accidents. A design of a data mining model for analysis and prediction of accidents rate in Nigeria was presented. In this study, we profiled traffic roads, differentiated the data set into pre- processing and transforming data set; created the association rules; and post-processed the frequent accident item sets. The data mining function was used and data cleaned using feature selection.