Machine learning based analysis and prediction of college students' mental health during COVID-19 in India

ABSTRACT

The traditional education system is focused on peer-to-peer learning, which helps in engagement, interactivity, and building confidence in students. However, the COVID-19 pandemic has shifted the focus from the traditional education system to online learning. The radical change in the education system has increased the mental stress in students. The parameters that affect the mental health of students include anxiety, academic stress, difficulty in concentrating, sleeping pattern disorders, decreased social interaction, job fear, etc. In this work, we design a modular framework for predicting the students' mental health based on a set of questions. First, we create a questionnaire to assess and analyze the parameters associated with mental health among college students in the Indian context. Based on these parameters, we first survey 600 students from Indian universities. Then, we categorize the students into two groups high mental stress and low mental stress using k-means clustering. Using the labels identified by k-means and further validated by the students, we then apply different classification models to predict the class of students. The proposed framework is experimentally validated through the dataset created from the questionnaire. In addition, we also analyze students' responses to find the parameters of utmost concern to the students. Therefore, our proposed work can be used to find the mental stress level of students so that corrective actions can be taken. © 2022 Elsevier Inc. All rights reserved.