Early Signs as Effective Indicators of Long-term Procrastination using Machine Learning

Abstract

Procrastination is a widespread issue that has a detrimental impact on students’ academic outcomes and mental health. Thus, for successful intervention, it is vital to identify early signs that correlate with a greater risk of procrastination. Most early-detection procrastination research focuses on the analysis of student activity and interaction with course material. However, the analysis of student-specific traits, such as personality and learning approach, in relation to procrastination habits is less studied. We designed an experiment to analyze the relationship between behavioral factors, specifically a student’s personality and time perspective, and long-term procrastination habits using Machine Learning. Personality traits will be reported using the Big 5 Personality Trait test, a survey containing 50 items that measure one’s levels of extraversion, agreeableness, openness, conscientiousness, and neuroticism. Time perspective will be reported using Zimbardo's Time Perspective Inventory, a survey containing 56 items that compute one’s score for each time perspective type – past negative, past positive, present fatalistic, present hedonistic, and future. Both surveys will be distributed to students at the beginning of a course for self-evaluation. Standard correlation and regression analysis will be conducted on this data to determine the relationship between these early signs and procrastination habits based on student submission times. Then, back propagation with perceptrons will be used to build a predictive neural network model based on these early signs to identify at-risk students. An IRB has been submitted to administer this study on a population of college students taking a core undergraduate course in Fall 2024.