Classification of Psychosomatic’s Symptoms of Depression: Iliou Versus PCA Preprocessing Methods

In this paper, we propose a novel data preprocessing method in order to facilitate the prediction performance of machine learning algorithms applied on datasets derived from mental patients. In this study, 136 questionnaires were distributed to mental patients - students with psychosomatic problems who were asked to volunteer at the University of Patras Specialty Health Service. The precision of the machine learning methods has to be very high for patients with this kind of issues, in order to achieve the sooner the possible the appropriate treatment. In our research, we used ILIOU data preprocessing method in order to enhance classification techniques for psychosomatic symptoms (i.e., depression). Firstly, we transformed the initial dataset with Principal Component Analysis and ILIOU data preprocessing methods, respectively. Afterwards, for the classification purpose we used seven machine learning classification algorithms with 10-fold cross validation method. According to the classification results, ILIOU preprocessing method led to a classification accuracy of 100% which is suitable for classification and prediction of psychosomatic symptoms.