Learning Student Stress Intention: An Interdisciplinary Analysis of Psychological, Environmental, and Academic Contributors using Machine Learning

Abstract

The research work focuses into all of the factors which contribute to student stress intention, through a dataset with 20 features spanning psychological, physiological, environmental, academic, and social domains. The research utilizes statistical tests and machine learning models such as Decision Tree, Random Forest, K Neighbors, and Gaussian Naive Bayes to find major stress predictors and evaluate model performance. Key findings show important relationships between every attribute and stress levels, with Gaussian Naive Bayes having the highest test accuracy. Feature importance analysis and dividing with Variational Autoencoders (VAE) and K-means reveal distinctive stress profiles, particularly in clusters 1 and 4, which might profit from centered mental health interventions. Association rule mining shows much stronger correlations between mental health signs like depression and anxiety. The study highlights the need for more data, ideal models, and customized interventions that better comprehend and manage student stress intention.