Behavioral Segmentation of Smartphone Users in Digital Society Using Unsupervised Clustering Techniques for Digital Engagement Analysis

The rapid growth of smartphone usage has significantly influenced digital behavior in modern societies. Understanding how individuals interact with smartphones is important for identifying patterns of digital engagement and potential risks associated with excessive usage. This study aims to discover smartphone usage behavior patterns in a digital society using clustering algorithms. A dataset containing 7,500 smartphone users with multiple behavioral variables was analyzed. An automated feature subset search strategy was applied to identify the most informative behavioral indicators, while clustering performance was evaluated using Silhouette Score and Davies–Bouldin Index. The optimal clustering configuration was obtained using the Birch algorithm with two clusters, supported by Principal Component Analysis retaining 85 percent of the dataset variance. The clustering model identified three key behavioral indicators that differentiate smartphone users, namely daily screen time, the proportion of social media usage relative to total screen time, and notification frequency per hour. The results revealed two distinct behavioral groups consisting of moderate engagement users and high digital engagement users. The high engagement cluster demonstrated significantly longer screen time, increased weekend smartphone usage, and a higher proportion of users labeled as addicted. These findings indicate that higher digital engagement intensity is strongly associated with smartphone addiction behavior. The study demonstrates the effectiveness of unsupervised machine learning in uncovering behavioral structures in digital environments and provides insights that may support future research on digital behavior, smartphone addiction, and digital well-being in contemporary digital societies.