Terrorist Group Classification of Historic Terrorist Attacks from The Global Terrorism Database

To conduct terrorist group classification of the terrorist attacks performed by unknown terrorist groups, text-based feature generation and logistic regression are used on the Global Terrorism Database (GTD). The model was trained and tested using stratified cross-validation with 10 folds, and a traintest-split of 75/25, achieving a median accuracy of 88%. Additionally, miscellaneous exploratory analysis is performed to capture and interpret macrotrends and characteristics that we present in the labeled and unlabeled GTD attack data.