Psychological Model for Mapping and Prediction of Stress Among Students

Stress has become a major issue for human beings and especially students, impacting both their general well-being and academic performance. Over the years, studies have revealed that academic stress and other stressors, such as time management, impede the smooth going of students in achieving their optimal academic performance and well-being. The work is aimed at using machine learning techniques to map and predict students’ stress levels using a psychological evaluation model. Data were collected from students of McPherson University using both the Perceived Stress Scale (PSS-10) and the 50-item International Personality Item Pool (IPIP) questionnaire. The data collected were preprocessed and trained by a variety of machine learning techniques to develop a psychological assessment model. Nine machine learning algorithms, which include Naive Bayes, Random Forest, Decision Tree, Logistic Regression, Linear Discriminant Analysis, Multilayer Perception, Bagging, Support Vector Machine (SVM), and K-Nearest Neighbour, were evaluated to determine the best for the model. Performance evaluation of the developed model is done using precision, recall, F1 score, and accuracy as metrics. The result shows that Random Forest is the best-performing classifier in this study, though with a low percentage due to the presence of imbalances in the data and feature selection.