Analisis Perilaku Penggunaan Smartphone dan Prediksi Kualitas Tidur Menggunakan Metode Statistik dan Machine Learning

Abstract

The rapid growth of smartphone and social media usage has reshaped daily digital behavior and raised increasing concerns regarding its potential impact on sleep patterns. This study investigates the relationship between digital usage behavior, psychological factors, and sleep outcomes using an integrated data science approach. A publicly available Social Media Mental Health Indicators dataset from Kaggle was utilized, comprising 5,000 observations that capture screen time, social media activity, digital interactions, psychological conditions, and sleep duration. Data analysis was conducted through a structured pipeline involving data preprocessing, exploratory data analysis, clustering, and supervised machine learning for classification and regression tasks. Exploratory analysis indicates consistent negative associations between screen-related variables and sleep duration. Clustering analysis reveals distinct behavioral groups characterized by different levels of digital engagement and sleep patterns. Furthermore, Random Forest models demonstrate reliable performance in both sleep quality classification and sleep duration prediction, highlighting their effectiveness in modeling complex and non-linear relationships. Feature importance analysis identifies screen time, social media intensity, and negative digital interactions as dominant contributors to sleep-related outcomes. These findings emphasize the value of combining statistical exploration and machine learning techniques to obtain a comprehensive understanding of how digital behavior relates to sleep, providing empirical support for data-driven evaluation of healthier digital habits.