Cluster Analysis of Pakistani Terrorism Events to Support Counterterrorism

Abstract

The frequency of terrorist events in Pakistan has increased considerably in the past several years. These events are frequent and not random, making it important to identify useful patterns in their occurrences to assist counterterrorism organizations. In this paper, we conducted such an analytical activity for the first time in Pakistan. We acquired data of terrorist events from reliable online sources and applied data preprocessing techniques followed by cluster analysis. Based on statistical correlation, we discovered clusters over the following combinations: (1) “Event of Terrorism—Target of Terrorism”; and (2) “Event of Terrorism—Method of Terrorism”. A more significant clustering is one which groups distinct combinations into separate clusters. We analyzed these clusters along three dimensions: (1) Annually for the time period 1988–2012; (2) for each Pakistani province; and (3) for different types of terrorist events. We also proposed a statistic for gauging the intensity of terrorism and analyzed it along the same three dimensions. Our results were extensive, but generally indicated significant Event–Target and Event–Method clusters, as well as increasing and decreasing trends in terrorism intensity. These can assist counterterrorism authorities in thwarting future attacks and arresting the responsible criminals.