Mobile Game Addiction Levels Among University Students Using Machine Learning

Abstract

Using mobile game addiction as a research focus, this case study examines how this digital disorder affects the academic performance and social life of university students. The study objectives include applying machine learning analysis for predicting the degree of mobile game addiction and creating a prediction model that will facilitate early intervention. One method used in this study was a cross-sectional survey of 600 university students on 17 variables: gaming tendency and several mood and social consequences. The obtained dataset underwent some preprocessing and encoding in order to be tested for different ml algorithms. Out of all the models, Random Forest Classifier achieved the highest accuracy of 96.45 % and Gradient Boosting Classifier Test set accuracy was 96.04% and for Decision Tree Classifier was 95.04%. Logistic Regression and GaussianNB had the lowest ranking, scoring 79.58% and 74.37%, respectively. Originally, the study result showed that feature selection and data preprocessing had a dramatic impact on model performance. Dependents is well classified with the Random Forest model and the wrongly classified addiction levels are demographically misclassified. Based on the study, it is critical to note that the various machine learning techniques can assist in identifying learners who require assistance concerning possible detrimental effects on their mental wellbeing. The subsequent research should recruit more participants and from a diverse population, while developing more intricate models for the improved prediction of depression risks.