Physiological and Motion Data to Extract Common Features of Procrastination Personas

Human beings are generally equipped with the trait of seeking escape from difficulties. Procrastination is a widespread problem in society. Especially, for educational settings, learner procrastination is a problem to be avoided. In this study, a new approach to finite procrastination traits is proposed that uses values from body sensor data. The proposed method creates a procrastination persona based on the relationship between the subject's motivation and their movements and physiological responses. The proposed method uses Nonnegative Matrix Factorization (NMF) to classify the collected data into clusters using an unsupervised machine learning model. The results of the experiment showed that subjects are divided into an average procrastination persona, a low procrastination persona, and a high procrastination persona. The need for intervention is lower for learners whose movement is less in the trunk and greater in both hands. Learners with low electrodermal activity and high heart rate require particularly active intervention. It is discovered that it is important to calm the heart rate and move the patient into a state of relaxation when intervening.