Prediction of Depression among Women Using Random Oversampling and Random Forest
Lee Ker Xin, Nuraini binti Abdul Rashid
Abstract
Mental Health is one of the significant issues affecting lives, especially during the COVID-19 pandemic. One of the many mental health problems is Depression. According to the latest Malaysian National Health and Morbidity Survey, around 30% of adults were affected by mental health sometimes in their lives. Machine Learning plays an essential role in uncovering knowledge hidden in medical data; hence, medical prediction can classify and predict illness. However, sometimes medical data can be imbalanced, thus affecting the accuracy of classifiers. Therefore, sampling can correct the balancing before classification. In this research, we employed SMOTE as a random oversampling method and Random Forest as a classification to accurately predict Depression among women. The generalized prediction model can be used to screen depression among women in Malaysia effectively.