Patterns of Terror: A Comparative Predictive Model

Abstract

This study aims to develop an intelligent model for counter-terrorism operations using a dataset of 50,064 entries from the Armed Conflict Location and Event Data (ACLED) covering terrorism activities from 1997 to 2022. The entropy-based feature engineering approach, utilizing information gain, was employed for its robustness to noisy data and ability to prioritize the most predictive features. Three machine learning classifiers—REPtrree, AdaBoost, and Meta-Stacking—were compared, with REPtree achieving the highest accuracy of 96.81% in predicting terrorist incidents. This study contributes to enhancing national security by leveraging machine learning techniques to improve counter-terrorism efforts.