DEVELOPING SPATIO-TEMPORAL DYNAMIC CLUSTERING ALGORITHMS FOR IDENTIFYING CRIME HOT SPOTS IN KUWAIT | ||||
JES. Journal of Engineering Sciences | ||||
Article 1, Volume 43, No 1, January and February 2015, Page 1-15 PDF (1.26 MB) | ||||
Document Type: Research Paper | ||||
DOI: 10.21608/jesaun.2015.111010 | ||||
View on SCiNiTO | ||||
Authors | ||||
Taysir H. A. Soliman 1; Khulood Al Ommar1; Youssef B. Mahdy2 | ||||
1Information Systems Dept., Faculty of Computers and Information, Assiut University, Egypt | ||||
2Computer Science Dept., Faculty of Computers and Information, Assiut University, Egypt | ||||
Abstract | ||||
As crime rates are increasing worldwide, crime mining requires more efficient algorithms that can handle current situations. Identifying crime hot spot areas via clustering spatio-temporal data is an emerging research area. In this paper, dynamic clustering algorithms for spatio-temporal crime data are proposed to detect hot crime spots in Kuwait. Kuwait governorates are taken as case study: the capital, Hawalli, Al-Ahmady, Al-Jahra, Al-Farawaniya, and Mubarak Al-kebeer. In addition, different crime types are considered: act of discharge and humiliation, adultery, aggravated assault, bribery, counter fitting, drugs, embezzlement, fight or resist employee on job, forging of official documents, weapon, robbery and attempted robbery, suicide and attempted suicide, and bank theft. Applying Random subspace classification to those clustered data, 98% accuracy and 99.4% ROC are obtained, having precision (98.7%), recall (98.4%), and F1 (98.28%). | ||||
Keywords | ||||
Spatio-temporal data mining; hot spot detection; intelligent crime mining; random subspace classification; and clustering | ||||
Statistics Article View: 78 PDF Download: 331 |
||||