01588182

Component

The recent advancements in the field of data mining have made vast progress in extracting new information and hidden patterns from large datasets which are often overlooked by traditional statistical approaches. Rather than confirming the existing hypothesis obtained from years of evidence-based statistical analysis, these methods focus on searching for new and interesting hypothesis which were previously unobserved. Road safety researchers working with the crash data from developed world have seen encouraging success in obtaining new insight into crash mechanism using data mining methods. However, it is unclear how these advance methods will perform on datasets from the developing countries which are prone to quality and under reporting issues. In this manuscript, data mining methods have been applied on the road traffic crash data of Bangladesh containing 14,462 cases from 2006-2010. The dataset was initially divided into pedestrian, single vehicle, double vehicle and multi-vehicle crashes and for each of these subsets, Hierarchical Clustering was employed to identify hazardous clusters. Next, Random Forest facilitated identifying important variables explaining each of these clusters. Finally, Classification and Regression Trees were generated for each of these clusters using the variables selected by Random Forest to understand crash mechanisms. The results could identify several interesting relationships and patterns which went past the insight gained through conducting temporal analysis and presenting descriptive statistics on selected factors with pre-defined vulnerable groups mainly inspired from the experience of the developed world.

This paper was sponsored by TRB committee ABE90 Standing Committee on Transportation in the Developing Countries.

01584066

16-1241

English

21p

2016

Transportation Research Board 95th Annual Meeting

Web

Figures; References (35) ; Tables

Crash analysis; Crash causes; Crash data; Data analysis; Data mining

Bangladesh

Data and Information Technology; Highways; Safety and Human Factors; I81: Accident Statistics

Transportation Research Board Annual Meeting 2016 Paper #16-1241

TRIS, TRB, ATRI

Jan 12 2016 4:33PM