An Evaluation of Hybrid Machine Learning Classifier Models for Identification of Terrorist Groups in the aftermath of an Attack

Abstract

The urgency of responding to a terrorist attack and the subsequent nature of analysis required to identify the terrorist group involved in an attack demands that the performance of the machine learning classifiers yield highly accurate outcomes. In order to improve the performance of machine learning classifiers, hybrid machine learning algorithms are used with the goal of improving the accuracy. The aim of the study was to build and evaluate hybrid classifier models for identification of terrorist groups. The research specifically sought to: build base classifiers (Naïve Bayes, K-Nearest Neighbor, Decision Trees, Support Vector Machines and Multi-Layer Perceptron); build hybrid classifier models from a combination of the base classifiers; and compare the performance of the hybrid and base classifiers. The study adopted an experimental research method using WEKA tool for data mining and real-world terrorist datasets for the period 1999-2017 for Sub-Saharan Africa region from the Global Terrorism Database. WEKA supervised filter Ranker was used to select 6 attributes of data and 784 records. The classifiers were evaluated using 10-fold cross validation. The study established that the optimal performance for all the classifiers was realized with a more balanced class at a resample rate of 1000%. The study concludes that hybrid classifiers perform better than base classifiers, and the best performing model was a hybrid combination of KNN, and DT. The study provides insights on the performance of hybrid machine learning classifiers and lays a foundation for further research in hybrid machine learning approaches.