Machine Learning Framework for Accurate Mental Stress Detection in University Students Using Random Forest Classification

The rapid increase in university students' mental stress requires a highly efficient and automated method of early detection for timely intervention. Existing methods struggle with processing large textual data, exhibit poor scalability, and fail to capture contextual meaning, leading to inaccurate stress classification. To overcome these shortcomings, in this study, a machine learning framework based on Random Forest for prediction and Word2Vec for text feature extraction was introduced, guaranteeing a strong representation of text as well as improved predictive performance. The dataset includes text posts from different online forums, which are preprocessed to eliminate inconsistencies, normalize the text, and extract useful features. The designed approach outperformed the standard approaches with an accuracy of 99%, precision of 96%, recall of 92 %, and F1-score of 94 %, compared to standard approaches such as Naïve Bayes, KNN, Decision Tree, and SVM. The empirical findings affirm the trustworthiness of the model in recognizing mental stress, marking its prospect for practical use in mental health observation and intervention.