Discovery of Association Rules in National Violent Death Data Using Optimization of Number of Attributes
Seung-Hyun Kim, Craig Dunham, Suryo Muljono, Albert Lee, Taehyung Wang · 2009
Over the last several years, data mining has been successfully used in various areas such as finance, security, telecommunication, science, retail industry, marketing, and Web. In this paper, we explain data mining application to crime analysis. Particularly, this work explains a set of procedures to find optimized number of attributes of the National Violent Death Reporting Database to predict types of violent death that has occurred. We also describe a set of activities of data mining we employed, including data preprocessing, data cleaning, data integration, data discretization, entropy, and information gain. As a result, we have successfully discovered interesting association rules that could be used by law enforcement and government agencies to help prevent violent deaths.