Physiological Signals based Student Grades Prediction using Machine Learning

Amidst today’s fast-paced world, the escalating levels of stress among students have become a significant concern, and it appears to be inversely related to their academic performance. There have been several types of research on the link between stress and grade prediction. Our study aims to leverage the available dataset about the stress level in such a way that can predict students’ grades to comment on the student’s performance. The dataset is collected using the wearable device during three exams containing electrodermal activity, skin temperature, and heart rate from 7680 student subjects. Our approach involves employing a variety of machine learning and deep learning algorithms, including Decision Trees Regressor, CNN Regressor, KNN Regressor, Random Forest Regressor, and Support Vector Regressor models. Through our investigation, we aim to demonstrate the potential of wearable devices in extracting meaningful insights from physiological signals to predict students’ academic outcomes. The challenges associated with wearable observations present unique opportunities for addressing and mitigating stress-related issues among students, and this study harnesses the power of a publicly available dataset to contribute to this important area of research.