Predicting Depression From Routine Survey Data using Machine Learning

Abstract

Around the world, 264 million individuals experience the ill effects of depression, which is one of the main sources of incapacity. A negative workplace can bring about several physical and medical issues and result in loss of efficiency. Studies have demonstrated that individuals are hesitant to reach out for help from mental health experts. This is primarily due to the stigma around mental health related issues. Taking into account the upsides of machine learning, we employed a variety of algorithms for predicting depression in people. The data used for this study is routine survey data-people were questioned about their home and workplace environment, family history of mental illness, etc. The algorithms employed for analysis were: K-Nearest Neighbours, Decision Tree, Multinomial Logistic Regression, Random Forest Classifier, Bagging, Boosting and Stacking. The results showed that best performance was obtained by using Boosting algorithm that gave the accuracy score of 81.75, followed by Random Forest Classifier at 81.22, and others. Depression is a prevalent issue and we hope that our findings will be helpful in its early prediction.