HYBRID CLASSIFICATION ALGORITHMS FOR TERRORISM PREDICTION in Middle East and North Africa

Abstract Machine learning methods used for prediction and decision support are of great concern nowadays. Methods for learning implicit, non-symbolic knowledge provide better predictive accuracy. But Methods for learning explicit, symbolic knowledge produce more comprehensible models. Hybrid machine learning models combine strengths of both knowledge representation of model types. In this research we compare predictive accuracy and comprehensibility of explicit, implicit, and hybrid machine learning models. This research based on predicting terrorist groups responsible of attacks in Middle East and North Africa from year 2009 up to 2013 by comparing various standard, ensemble, hybrid, and hybrid ensemble machine learning methods namely; Naïve Bayes, K-nearest neighbours, Decision Tree, Support Vector Machine; Hybrid Hoeffding Tree, Functional Tree, Hybrid Naïve Bayes with Decision Table, Classification via Clustering; Random Forests; and Stacking classifiers. Afterwards compare the results obtained from conducting the experiments according to four different performance measures. Experiments were carried out using real world data represented by Global terrorism Database (GTD) from National Consortium for the study of terrorism and Responses of Terrorism (START).