Mental health conditions such as depression and anxiety have a widespread prevalence rate across the globe, thus calling for reliable models that can facilitate accurate prediction. The current research aims to establish a Random Forest model to classify and predict depression and anxiety in terms of their severity levels based on the DASS-21 dataset. The data set comprises 21 questionnaire-based features corresponding to the emotional state of patients. These features undergo preprocessing based on missing value imputation, normalization, and encoding methods. The data is partitioned into training and testing data sets at a ratio of 80:20. Next, a Random Forest classifier is trained to classify patients into various levels of depression and anxiety. The experimental outcomes reveal that the developed approach exhibits high prediction accuracy with 97% accuracy, and the precision, recall, and F1 scores are equally good. The model's stability is confirmed using confusion matrix evaluation, where misclassification between severity levels is negligible. In addition, there is an analysis of the importance of the characteristics that influence the results of the predictions. This increases the visibility of the model and helps to understand which psychological factors are essential. It becomes clear that the use of the Random Forest algorithm for predicting mental disorders can be quite effective.

Random Forest Mental Health Prediction Depression Detection Anxiety Classification DASS-21 Dataset Machine Learning