Identifying High and Low Academic Result Holders through Smartphone Usage Data

Abstract

Nowadays, smartphones have become an inseparable part of students' life. Many previous studies have explored smartphone usage behavior in different contexts. However, to our best knowledge, smartphone usage behavior of the high and low academic result holders is very less studied. Thus, in the context of Bangladesh, using 7 days' actual smartphone usage data of high [N=32] and low [N=44] performers, we investigate the smartphone usage of these two groups. Our findings show that low performers are more focused on certain apps of the Launcher category whereas high performers are more focused on certain apps of the Video category. Moreover, we find that low performers' micro usage and review session number is statistically significantly (p<0.05) higher. Based on different smartphone usage data, our presented machine learning model classifies these two groups of students with 73.33% accuracy. Thus, these findings suggest that high and low performers can be identified through smartphone usage data.