Predictive Modeling of the Impact of Smartphone Addiction on Students’ Academic Performance Using Machine Learning

ABSTRACT Objectives: The goal of this study was to use machine learning techniques to create and validate predictive models for detecting smartphone addiction. The study sought to find significant aspects linked to smartphone addiction and assess the models’ capacity to correctly recognize those at risk by examining a mix of data such as behavioral, psychological, and demographic. Methods: Five hundred participants between the ages of 16 and 45 made up the dataset. Data such as self-reported smartphone usage habits, Smartphone Addiction Scale (SAS) scores, and demographics were all included in this study. Recursive Feature Elimination (RFE) and other feature selection approaches were used to determine the important predictors of smartphone addiction. Predictive models were then built using machine learning techniques like Random Forest, Gradient Boosting, and Logistic Regression. The dataset was split into subsets for training (70%) and testing (30%), for developing and assessing the model. Key metrics like accuracy, precision, recall, and the F1-score were used to evaluate the model's performance. Findings: With an accurate record of 91.2%, precision of 88.7%, recall of 90.5%, and F1-score of 89.6%, the Gradient Boosting machine learning model outperformed the other techniques. Daily screen time, app usage frequency, sleep disturbance from smartphone use, and psychological traits like impulsivity and anxiety were among the major indicators found. Novelty: By combining behavioral data with sophisticated and intricate machine learning models, this study presents a novel and notable method for accurately predicting smartphone addiction. In contrast to other research, this study focuses on using explainable AI methods to derive useful insights, which could enhance the interpretability of predictive models for more future purposes.