Data-Driven Study of UK Terrorism: A K-prototypes Clustering Analysis of the UK's Terrorist Incidents

Terrorist attacks have become hard to predict and even harder to analyse in the last few decades. The constant morphing of terrorist typology combined with the myriad of groups and goals make it extremely difficult to pinpoint when and where a new attack might take place. This study aims to address the missing link between terrorist attacks in the UK and any hidden patterns that might be revealed through the use of K-prototypes clustering, preferred for its ability to analyse both categorical and non-categorical data simultaneously. This quality makes K-prototypes an optimal algorithm when dealing with the diverse nature of terrorist incident records. By creating a specialized subset extracted from the Global Terrorism Database (GTD), this research focuses specifically on incidents comprised of both categorical and non-categorical features filtered by the preferred type of attack (bombing/explosion)Through this analysis, distinct clusters of bombing/explosion attacks favouring the use of some kind of explosive device emerge: successful attacks on military targets in London, successful attacks on police in Belfast, and unsuccessful attacks on private citizens, also in Belfast. Nevertheless, all these various attacks result in a low number of casualties and injuries, suggesting limited effects or a propensity for targeting properties over people. Moreover, there is no propensity for suicide terrorist attacks. This study also highlights the prevalence of attacks in Northern Ireland, indicating a possible need for enhanced counter-terrorism security measures in this region. Ultimately, the research explores new ways to identify previously unseen patterns of terrorist behaviour and highlight persistent threats. It also underscores the importance of parameter selection, while acknowledging limitations such as data bias and geographical focus. The practical implications vary from targeted security enhancements, improved intelligence gathering, or even creating more comprehensive strategies to prevent future attacks.