Terrorism is a complex, dynamic and continuously evolving phenomenon. In past few decades, Pakistan has seen an increased number of such incidents across the country. Currently, these terrorist incidents are highly unpredictable which gives an edge to the terrorist groups to attack by surprise. Main reason for this unpredictability is due to lack of real time application which predicts such incidents based upon risk models and prediction algorithms.
In this project, we developed a model that carries out comprehensive trend/ cluster analysis and further predict potential threats including patterns and regions. Patterns are identified from the past occurrences covering location, groups involved, victims and historical incidents in the region. This research focuses on extraction of these patterns from the historical data to formulate clusters of incidents and subsequently form a risk model for incident prediction.
We have used following datasets:
- Global Terrorism Database (www.start.umd.edu/gtd)
- Wikipedia List of Terrorist Incidents (en.wikipedia.org/wiki/List_of_terrorist_incidents_in_2017)