Detecting Depression: Employing Natural Language Processing and Random Forests

Developed an NLP-based depression-detection model using Random Forest classifiers, achieving a 93.24% F1-score and 93.21% accuracy on a curated dataset of 989 Reddit posts (Depression / Control).

Built the data-collection and labeling strategy and an optimized pipeline using TF-IDF features, SMOTE balancing, and Grid-Search cross-validation.