Comparative Analysis of Mental Health disorder in Higher Education Students using Predictive Algorithms

Mental health problem is the biggest problem in the current world. The problem in the case of higher education students is one of the challenging factors. To solve the issue the concern is to collect all the factors then analyses and then implement the methods to find out the cause After getting the full and effective cause, it is also very challenging to implement. But the most effective factor is how the problem can be diagnosed. There are different emerging techniques are developing day to day. But the most effective approaches are to implement machine learning applications or algorithms to detect the solution in an effective manner. This study is based upon the classification tasks. The mental health problems are different types like stress, anxiety or depression etc. The different classification algorithms are implemented over these problems and suitable algorithm is picked as per the performance measurement. The performance evaluation is also done by taking the factors associated with the particular mental health problem. In the study 4 classifications are used as Decision Tree (DT), Logistic Regression (LR), Support Vector Machine (SVM) and Neive Bayes (NB). After successful implementation, different algorithms show different performances. For stress detection Decision Tree algorithm’s performance came out as best, while for Anxiety level detection Logistic regression is the out performer. In same way support vector machine works good for depression level calculation. The result shows that machine learning algorithms are best if the input data is trained, tested. Overall, the outcome shows that the methods or machine learning algorithms which are implemented have significant influences. The outcomes can be implemented over the higher education students to detect the mental health problem status. The overall result shows that for stress detection Decision Tree gives 94.4%, Support Vector machine gives result of 91.2% for depression level calculation. For Anxiety level detection, Logistic regression shows 88.76% which is best over all other algorithms.