Analyzing Social Media Addiction Among University Students Using Machine Learning and Behavioral Data

The rapid growth of social media usage has significantly influenced the daily behavior of university students, raising concerns about the potential development of social media addiction. This study aims to analyze behavioral patterns related to social media addiction among students and to evaluate the effectiveness of machine learning techniques in predicting addiction levels. The dataset used in this research consists of 705 student records containing demographic characteristics, behavioral variables, and social media usage attributes. Several preprocessing steps were conducted before applying machine learning models to classify students into low and high addiction categories. Two classification models, Logistic Regression and Gradient Boosting, were implemented and evaluated using accuracy, F1 score, and ROC AUC metrics. The results indicate that the Gradient Boosting model achieved the best predictive performance, demonstrating strong capability in identifying patterns associated with social media addiction. Feature importance analysis further reveals that behavioral factors play a more substantial role than demographic characteristics in predicting addiction levels. In particular, perceived academic impact and sleep duration emerge as the most influential predictors. These findings suggest that social media addiction among students is closely related to behavioral patterns in daily digital engagement. Overall, this study highlights the potential of machine learning approaches to support the analysis of digital behavior and to provide insights into factors associated with problematic social media use among university students.