Predicting Procrastination in College Students Using Machine Learning: A Comparative Analysis of Models – A Preliminary Experiment


  • ARNAV MATHUR Westfield High School, Chantilly, VA
  • Mihai Boicu Information Sciences and Technology, George Mason University, Fairfax, VA



Procrastination is a prevalent issue among college students that can negatively impact academic performance and overall well-being. Several machine learning techniques were explored to predict procrastination tendencies in college students based on daily behavioral patterns and task characteristics (assignment type, grades, and submission time for each attempt). A procrastination index was created based on late submission frequency and grade improvement frequency, with a higher index indicating increased procrastination behavior. Three machine learning models are compared: Random Forest, Support Vector Regression, and Neural Network, to identify the most effective approach in predicting the index. Using the Random Forest Regression model, an R-squared value (accuracy) of 0.873 was achieved, indicating a strong ability to predict procrastination tendencies based on the given features. The Support Vector Regression showed an R-squared value of 0.609, and the Neural Network achieved an R-squared value of 0.840. The findings demonstrate that machine learning models can predict procrastination tendencies in college students. This research is a preliminary step to eventually better understanding procrastination behaviors and providing early intervention and support for college students facing procrastination challenges. 





College of Engineering and Computing: Department of Information Sciences and Technology