부산대학교 도서관

Over the past few decades, many people are suffering from psychological health issues such as anxiety, depression, and stress. It is crucial to detect mental health condition on a timely basis and cure before it turns into a severe problem. In this paper, we analyze the performance of Machine learning (ML) algorithms to predict severity levels of depression, anxiety, and stress (DAS). Severity levels can be measured by the Depression, Anxiety, and Stress Scale (DASS) which consists of a set of questionnaires (DASS42). The intention of this research is to enhance and compare the performance of two different ML algorithms namely Support Vector Machine (SVM) and Logistic Regression (LR) based on classification accuracy with other algorithms. After parameter tuning SVM attains a classification accuracy of 97.35%, 97.49%, and 97.20% for Depression, Anxiety, and Stress. LR attains a classification accuracy of 98.15%, 98.05%, and 98.45% for Depression, Anxiety, and Stress dataset respectively. Parameter tuning is done for SVM and LR to obtain better accuracy with the best suitable parameters, and the results revealed that LR achieved better performance in terms of accuracy compared to SVM.