Researchers from the University of Southern California Viterbi Information Sciences Institute have mined sensor data collected from volunteers to discover hidden trends in well-being, academic performance, and behavior.
They used publicly available data from the StudentLife application, which monitored readings from smartphone sensors carried by 48 Dartmouth College student volunteers over 10 weeks in 2013.
The team employed computational models to split students into two groups based on their behavioral patterns. Using non-negative tensor factorization, they identified clusters of individuals who showed higher academic performance, as well as those who frequently engage in leisure activities.
They observed that students with higher grade point averages spent a greater amount of time at the library, less time relaxing, and were generally asleep before midnight. In contrast, students with poorer grades engaged in more conversations between midnight and 6 a.m., and spent more time in the dark during morning hours.
From USC Viterbi News
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found