The Student Mental Health Pattern Using Clustering and Classification Approaches

Abstract

Students mental health is a key factor in their academic and social development.
However, the patterns and factors that influence mental health in college students are
still not fully understood. This study utilizes machine learning-based clustering and
classification techniques to identify hidden patterns in college students’ mental health
data, focusing on social and demographic factors. Using the K-Means algorithm for
clustering and Random Forest for classification, we group college students based on
their mental health conditions and analyze the associations between variables such as
age, marital status, anxiety, and medical history. The process begins with data
exploration, followed by data cleaning and feature transformation to ensure optimal
input quality. In the clustering stage, we find three main groups of college students with
different mental health patterns, which are then used as the basis for a classification
model. A Random Forest model is built to predict potential mental disorders, such as
depression and anxiety, by identifying the features that have the most influence on the
prediction results. The model evaluation shows significant performance with adequate
accuracy, where the importance of social factors such as marital status and history of
visits to medical professionals is clearly revealed. The results of this study not only offer
important insights into students’ mental health patterns, but also provide
recommendations for university policies in creating an environment that supports
students’ mental well-being. This combined approach of clustering and classification
opens up new opportunities in the application of machine learning for more precise and
data-driven mental health analysis.