Abstract:
This study aims to develop a predictive model for smartphone addiction among
university students in Bangladesh using machine learning techniques. The research
encompasses data collection through a comprehensive survey, which included
demographic information and responses to 25 questions designed to assess smartphone
usage patterns and behaviors. A weighted composite score was calculated for each
respondent to represent their level of addiction, which was then normalized for
standardization. Various machine learning models were trained and evaluated,
including logistic regression, decision trees, and support vector machines, among
others. The logistic regression model demonstrated the highest accuracy of 97% in
predicting smartphone addiction. The study also identified key factors associated with
smartphone addiction, such as loneliness, depression, and poor sleep quality.
Additionally, the research found that smartphone addiction had negative impacts on
students' academic performance and social interactions. The findings of this study
provide valuable insights for university administrators and policymakers to develop
targeted interventions and support services to address the growing issue of smartphone
addiction among university students in Bangladesh. The predictive model can be further
refined and integrated into student well-being programs to identify and support those at
high risk of addiction.
