Mental Health Prediction Among Unemployed Graduates Using Machine Learning Approach: BD Perspective

Abstract

According to emerging advancements, university graduates looking for work are more likely to experience common psychological illnesses including obsessive thinking, panic attacks, nervousness, or distress. Nevertheless, Bangladesh has still not looked into the difficulties with mental health faced by unemployed graduates. We have developed a model that employs a typical psychological screening using machine learning algorithms to identify the various stages of certain psychological illnesses among unemployed Bangladeshi graduates in order to identify these problems at a young age. In order to effectively assist unemployed graduates, our focus was to obtain accurate and reliable sets for identifying trends such as age, gender, the causation of sadness, the data of behavioral change, as well as several other things. The questionnaires that have been obtained throughout the duration of this study has been of considerable assistance in terms of forecasting depression and giving counsel. On the real datasets for predicting depression, we applied seven different types of ML algorithms in our suggested model: Random Forest Classifier, Gaussian Naive Bayes, K-Neighbors Classifier, Logistic Regression, Support Vector Machine, Gradient Boosting Classifier, AdaBoost. Our model's outcomes show that Random Forest, Logistic Regression, Gradient Boosting Classifier performed admirably overall with the accuracy of 80.6%. Remaining algorithms outperformed the others we used by a significant margin. We have a clear focus to add to this conversation with the recommendation model that we developed as part of this study. Using the assistance of Machine Learning by monitoring and analyzing those behaviors, we believe that we will be capable of assisting them in the near future with website or mobile applications as they can aware themselves.