Stress Analysis for Students in Online Classes

This paper aims to identify the stress levels of students in Massive Open Online Courses(MOOCs). Research shows that there is a lack of sentiment analysis for online classes and hence a higher attrition rate. We thus aim to help instructors identify the stressed students. Using student posts from online platform “Piazza” as input, we perform various stress detection analysis methods like Naive Bayes, ANEW, VADER and SentiWords. These stressed posts from each method are extracted to compare accuracy with baseline dataset. This research provides unique solutions to detect the student sentiment in formal environment which can help reduce stress and improve the students' overall performance.