Utilization of machine learning techniques to predict the relationship between depression and university students’ life

Depression is a common psychological disorder that affects an individual's feelings, thinking, and behavior. An intense feeling of sadness and loss of pleasure and interest in daily life characterizes it. Depression can affect a person's ability to perform daily duties and may make it difficult for him to understand his psychological state. Symptoms of depression include constant sadness, not enjoying daily life, sleep disorders, low energy, fatigue, and in advanced cases frequent thoughts of death or suicide. Concerning college students, studies show that it is between 20% and 30%, and in some poor developing countries, it may reach 50% or 60%. Therefore, this paper aims to identify the most important and relevant features that greatly facilitate early prediction of depression in students. Also, one of the main objectives of this paper is to determine the best classifier that could be used to predict the existence of depression among 10 models that belong to different learning strategies and using several evaluation metrics. Moreover, the paper aims to determine the best feature selection that can determine the best feature among all features. The results showed that Gradient Boosting and Logistic Regression are the best classification models. Among the feature selection methods, Gini Ratio and Chi-Square (χ²) give better results in general than other feature selection methods, and these results are good for mental health professionals, which helps in diagnosing whether the student has depression or not.