Examining the factors that predict the likelihood of the success of a terrorist attack, and severity from a machine learning and regression lens
Citation & Export
Hide
Simple citation
Tarakji, Mohamed.
Examining the factors that predict the likelihood of the success of a terrorist attack, and severity from a machine learning and regression lens. Retrieved from
https://doi.org/doi:10.7282/t3-b2j5-v026Export
Description
TitleExamining the factors that predict the likelihood of the success of a terrorist attack, and severity from a machine learning and regression lens
Date Created2021
Other Date2021-05 (degree)
Extent1 online resource (xii, 107 pages)
DescriptionTerrorism and the proactive and reactive policies enacted to counter further terrorism incidents have consequential effects on both a microeconomic and macroeconomic level. Although terrorism has existed prior to the modernization of the global economy, we are witnessing a recent increase in the aggregate number of attacks, as well as in the severity of the attacks, in recent decades. This rise in terrorism and its impacts can be attributed to the rise of religious Islamic terrorism, and the shift in motivations of terrorist organizations since the post-cold war era. Furthermore, terrorist organizations are adapting to counter-terrorism efforts as they shift their activities to regions that are less secure and adapt their methodologies of attack to increase their likelihood of success. As governments and global organizations continue to increase their spending in efforts to counter terrorism, if efforts are to be effective and efficient it requires a systematic approach in considering the magnitude of impact of various methods of attack, the likelihood of success each method, and the expected risk reduction and implications of implementing security measures.
The political and social science fields have yet to be able to reach an agreed upon definition of terrorism, or determine the underlying factors which are both controversial topics in the field of terrorism. This lack of a standardized definition constitutes an obstacle to effectively implement proactive policies that reduce individuals’ participation in terroristic activities. In the meanwhile, the field of applied economics can contribute to the effective implementation of defensive strategies by lending its statistical and econometric tools to the research in this field. Furthermore, as data on this topic continues to expand and with the availability and maintenance of large datasets by governmental agencies, machine learning is a useful tool in researching this issue as it is largely data driven and does not assume any functionality of the relationships. Machine learning is not to a replacement for classical econometric methodology but can instead enhance these models by being better suited for predicting out of sample observations when utilizing large datasets. By merging these two field together, economics and machine learning, we can better predict the likelihood of success of a terrorist attacks and its magnitude while not sacrificing the ease of interpretability provided by regression modeling techniques.
Our research contributes to the study of terrorism by examining the factors that predict the likelihood of success of a terrorist attack as well as the severity by utilizing a combination of machine learning algorithmic approaches and economic regression modeling techniques. To the best of our knowledge, this approach has yet to be implemented to the research in this field, and thus constitutes a novel contribution to the literature. From our results we observe that the decade, region, method of attack, weapon utilized in the attack, attack target, whether the attack was transnational or domestic, whether the perpetrator/s committed suicide, and the duration of the attacks are statistically significant predictors as to whether the attack was successful or not. Furthermore, these factors are also significant in regard to measuring the magnitude of the severity of the attack, as measured by individuals killed and number of wounded casualties. Separate from our model that predicts the likelihood of the success of the attack, two separate models for each measurement of severity are also constructed throughout our research. Considering the lack of definition and agreement on the underlying factors that motivate individuals to participate in terrorism, our focus is on identifying the factors that predict the likelihood of the success and severity of a terrorist attack. By examining the factors and methodologies that contribute the highest to the predictive likelihood of the success a terrorist attack and that have the highest magnitude of impact, this study offers policy implications to the design of effective defensive policy measures and efficient resource allocation recommendations to policymakers and governmental agencies to further thwart terrorism efforts.
NoteM.S.
NoteIncludes bibliographical references
Genretheses, ETD graduate
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.
Statistical Profile