Predicting Social Media Addiction Using Data Mining Techniques: Development of a Web-Based Decision Support System

Abstract

The purpose of this study is to analyze high school students’ social media usage habits, develop a classification model to predict social media addiction risk, and design a web-based decision support system. Data were collected from 495 students enrolled in different types of high schools in Tokat city center, consisting of 55 attributes derived from a social media addiction scale and demographic information. Following the preprocessing stage, an information gain-based feature selection was applied, and 24 attributes with information gain values above the average threshold (0.156) were identified. Seven different classification algorithms were evaluated using 10-fold cross-validation, and the resulting models were compared in terms of precision, recall, F1-score, and accuracy. The highest performance was obtained with the SMO algorithm, yielding precision = 0.93, recall = 0.95 and F1-score = 0.95; the model classified social media addiction into low, medium, and high levels with an accuracy rate of 95.15%. The developed model was integrated into a web-based decision support system, providing school counselors with a practical tool for early detection and intervention processes. The study presents a web-based decision support system that supports intervention strategies in secondary education institutions by achieving high accuracy in addiction detection.