Estimating the Uniqueness of Linked Residential Burglaries. A Data Mining Approach

      Аннотация: There has been a consistent increase in residential burglary crimes in Sweden compared to past decade. Inspite of knowing the crimes being committed by few offenders, Law enforcement agencies were only able to solve 3-5 percent of crimes that took place in 2012. Abiding to crime analysis studies, they thought of investigating the possibility of linking crimes into crime series. Based on this research gap this book describes the approach for linking crimes(burglaries) into crime series based on their characteristics. An approach called Median crime and a measure called Uniqueness measure is developed for this purpose. A state-of-the-art algorithm is chosen for building a statistical model that is being used for linking burglaries into a series. A median crime is computed and uniqueness measure is calculated for each of these formulated series and known series that are legally verified by law bodies. The uniqueness measure of formulated series is compared to known series estimate the feasibility of using the formulated series for investigation by law bodies. This book serves as a purpose to know about linking residential burglaries and methods to compare entire crime series.

Ключевые слова: Data Mining, Logistic Regression, Machine Learning, Sweden, uniqueness, Residential burglaries, Median crime