Analysis of Mental Health using Machine Learn...

Analysis of Mental Health using Machine Learning Algorithms

Author Name : Sounak Nandi

Using the Mental Health Corpus dataset from Kaggle, comprising 27,977 annotated text entries, text preprocessing techniques such as data cleaning, tokenization, stop-word removal, and sentiment feature extraction are employed. Six machine learning algorithms—Random Forest, Support Vector Machine (SVM), Neural Network, XGBoost, K-Nearest Neighbour (KNN), and Logistic Regression—were trained and evaluated. SVM with a linear kernel emerged as the most effective, achieving 91.14% accuracy and a ROC-AUC score of 0.9698. Key features included terms related to depressive symptoms. This improves mental health prediction accuracy and enhancing screening methodologies