COMPARISON OF THE CLASSIFICATION PERFORMANCES OF CRIMINAL TENDENCIES OF SCHIZOPHRENIC PATIENTS BY ARTIFICIAL NEURAL NETWORKS AND SUPPORT VECTOR MACHINE

Abstract

In this study, a new approach based on Artificial Neural Networks (ANN) and Support Vector Machine (SVM) classifiers has been proposed in the determination of criminal tendency with biochemical data of schizophrenia patients. Classification was performed using the biochemical data of the offender and control group schizophrenic patients. The data were obtained from 100 schizophrenic inpatients in Elazig mental and Neurological Disorders Hospital. The biochemical data used for the examination and classification of the criminal tendencies of schizophrenic patients were Triglycerides, Total Cholesterol, High Density Lipoproteins (HDL), Low Density Lipoproteins (LDL), Very Low Density Lipoproteins (VLDL), Sex Hormone Binding Globulin (SHBG), Oestradiol, Free Testosterone, Total Testosterone, Ghrelin, Copper (Cu) and Zinc (Zn). Biochemical data were classified using ANN and SVM. All data were normalized to before classification. In addition, classifier results were evaluated using cross-validation method. As a result of the classification performed, 87% accuracy and 89% accuracy were achieved by ANN and SVM, respectively. In the determination of the criminal tendencies of schizophrenic patients using their biochemical data, SVM classifier performed a more effective classification compared to ANN classifier. According to classification results, it was seen that the biochemical data used could be useful features in the determination of the criminal tendencies of schizophrenic patients.