Criticality Trend Analysis Based on Different Types of Accidents using Data Mining Approach

Safety on roads and prevention of accidents are the prime concern of any highway system. Data mining is a source of retrieval of information for knowledge discovery approach. Many data mining methodologies have been applied to accident data in the recent past years. There is need to analyze the relationship between different factors related to accidents i.e. number of persons affected by fatal, minor, grievous, non-injury, road feature (ROF), road condition (ROC), cause of accident (CAU) and vehicle responsible (VR) according to daily, fortnightly, semi-fortnightly and monthly basis. The objective of this study is divided into three sub-objectives. The First sub-objective of this study is to divide number of accident dataset of National Highway sections of Karnataka state implemented by Project Implementation Unit i.e. PIU (Bangalore, Chitradurga, Dharwad, Gulbarga, Hospet and Mangalore) during January 2012 to January 2017 collected from NHAI (National Highway Authority of India) in homogeneous clusters using K-means clustering. The second sub-objective is to reflect the relationship between different factors i.e. a number of persons affected by fatal, minor, grievous, non-injury, CAU, ROC, ROF and VR using Apriori association rule. The last sub-objective is to perform temporal trend analysis for each cluster on the basis of rules generated by Association Rule Mining.