Particle Swarm Optimization Feature Selection for Violent Crime Classification

Abstract

Crime prevention is one of the important roles of the police system in any country. One of the components of crime prevention is crime rate classification. Thus, this study proposed a crime classification model by combining Artificial Neural Network (ANN) model and Particle swarm optimization (PSO) model. PSO is used as feature selection to select the significant features that affects the capability of ANN as classifier. This combination is expected to generate more accurate classification result with minimum error. To evaluate the performance of the proposed model, comparison with ANN model without PSO is carried out on the Communities and Crime dataset. The proposed model is found to produce better classification accuracy as compared to ANN model alone in classifying crime rates. Besides improving the classification accuracy, the proposed model has reduced the learning convergence time in training phase.