AdaBoost based hybrid concept drift detection approach for mental health prediction
Meenal Jain, Saksham Gupta, G Agarwal, Kritarth Bansal · 2023
One of the most crucial components of reducing the risk of serious mental disease is mental health prediction. Meanwhile, the department of public health may apply mental health prediction as a theoretical foundation to develop plans for behavioral therapies for health care workers. This study uses machine learning to make predictions about mental health. Conventional machine learning techniques, however, have difficulty providing reliable high accuracy values and generate a lot of false alarms. This is because the presence of concept drift in the continuous data. Concept drift is the term used to explain unforeseen changes in data features over time. Therefore, novel technique is needed to handle the presence of concept drift. In this study, we presented the Fixed Sliding Windowing (FSW) approach and studied its effects. In order to predict mental health, we employed the AdaBoost classifier, and we began retraining the model based on statistical tests. AdaBoost has been found to have a higher accuracy ratio after applying the proposed method.