Identifying Adolescent Behavioral Profiles Through K-Means Clustering
Based on Smartphone Usage, Mental Health, and Academic Performance
Dublin Core
Title
Identifying Adolescent Behavioral Profiles Through K-Means Clustering
Based on Smartphone Usage, Mental Health, and Academic Performance
Based on Smartphone Usage, Mental Health, and Academic Performance
Subject
Machine Learning, Adolescent, Smartphone Usage, Mental Health, K-Means Algorithm, Cluster Analysis, Addiction Patterns
Description
The pervasive integration of digital devices into students’ daily lives has profoundly shaped their learning habits and psychological well-being.
As technology becomes increasingly embedded in academic and personal routines, understanding the relationship between digital engagement,
mental health, and academic outcomes is vital for developing effective student-support and intervention frameworks in higher education. This
study seeks to uncover behavioral patterns among college students by examining the interconnections between smartphone usage, mental health
indicators, and academic performance through a data-driven machine learning approach. Utilizing the K-Means clustering algorithm, students
were categorized into distinct behavioral profiles derived from eight core features: daily screen time, sleep duration, grade performance, exercise
frequency, anxiety level, depression level, self-confidence, and screen exposure before sleep. A dataset comprising 3,000 entries was preprocessed
through normalization and analyzed within the Knowledge Discovery in Databases (KDD) framework to ensure structured and reliable data
processing. The Elbow Method identified four optimal clusters, each reflecting unique behavioral characteristics. Cluster 1 represented wellbalanced students with stable academic and emotional states; Cluster 2 included high-achieving yet anxious individuals; Cluster 3 captured those
exhibiting excessive digital engagement and psychological distress; and Cluster 4 comprised moderately engaged students with lower selfconfidence. Visual representations, including bar and radar charts, were generated to illustrate inter-cluster variations and enhance interpretability
of behavioral distinctions. The findings reveal that digital usage patterns are closely linked to mental health and academic performance, suggesting
that excessive or unregulated device use can heighten emotional strain and academic inconsistency. These insights highlight the necessity of
personalized mental health initiatives and targeted digital literacy programs grounded in behavioral segmentation. Overall, the study demonstrates
the applicability of unsupervised machine learning for behavioral profiling and provides evidence-based recommendations for educators, mental
health practitioners, and policymakers seeking to foster balanced and healthy digital habits among students
As technology becomes increasingly embedded in academic and personal routines, understanding the relationship between digital engagement,
mental health, and academic outcomes is vital for developing effective student-support and intervention frameworks in higher education. This
study seeks to uncover behavioral patterns among college students by examining the interconnections between smartphone usage, mental health
indicators, and academic performance through a data-driven machine learning approach. Utilizing the K-Means clustering algorithm, students
were categorized into distinct behavioral profiles derived from eight core features: daily screen time, sleep duration, grade performance, exercise
frequency, anxiety level, depression level, self-confidence, and screen exposure before sleep. A dataset comprising 3,000 entries was preprocessed
through normalization and analyzed within the Knowledge Discovery in Databases (KDD) framework to ensure structured and reliable data
processing. The Elbow Method identified four optimal clusters, each reflecting unique behavioral characteristics. Cluster 1 represented wellbalanced students with stable academic and emotional states; Cluster 2 included high-achieving yet anxious individuals; Cluster 3 captured those
exhibiting excessive digital engagement and psychological distress; and Cluster 4 comprised moderately engaged students with lower selfconfidence. Visual representations, including bar and radar charts, were generated to illustrate inter-cluster variations and enhance interpretability
of behavioral distinctions. The findings reveal that digital usage patterns are closely linked to mental health and academic performance, suggesting
that excessive or unregulated device use can heighten emotional strain and academic inconsistency. These insights highlight the necessity of
personalized mental health initiatives and targeted digital literacy programs grounded in behavioral segmentation. Overall, the study demonstrates
the applicability of unsupervised machine learning for behavioral profiling and provides evidence-based recommendations for educators, mental
health practitioners, and policymakers seeking to foster balanced and healthy digital habits among students
Creator
Dominic Dinand Aristo1,*
, Bhavana Srinivasan2
, Bhavana Srinivasan2
Source
https://ijiis.org/index.php/IJIIS/article/view/231/154
Publisher
Amikom Purwokerto University, Indonesia
Date
januari 2025
Contributor
Fajar bagus W
Format
PDF
Language
English
Type
Text
Files
Collection
Citation
Dominic Dinand Aristo1,*
, Bhavana Srinivasan2
, “Identifying Adolescent Behavioral Profiles Through K-Means Clustering
Based on Smartphone Usage, Mental Health, and Academic Performance,” Repository Horizon University Indonesia, accessed January 2, 2026, https://repository.horizon.ac.id/items/show/9722.
Based on Smartphone Usage, Mental Health, and Academic Performance,” Repository Horizon University Indonesia, accessed January 2, 2026, https://repository.horizon.ac.id/items/show/9722.