Internet Addiction Predictor: Applying Machine Learning in Psychology

Abstract

This work is an effort to exploit the unique ability of supervised machine learning model to explore how causal systems have an influence on Internet addiction disorder occurrence. The paper aims to pursue the possibilities of predicting Internet addiction based on a set of predictor variables. Here, the predictor variable set is selected such that there exists a strong relationship between the parameters considered to have an influence toward problematic Internet usage. Healthcare sector data always poses a challenge to the researchers in terms of unbalanced size of class representative samples available for the study. This kind of unbalanced dataset does affect the efficiency of the machine learning model to learn unbiased and to predict the unseen/test data accurately. Further, to this challenge, we propose to make a study on understanding the effect of an unbalanced dataset on the efficiency and performance of the machine learning model.