Depression is an important psychological problem that affects an extensive group of individuals all around the world, including an assortment of folks today. Depression is inclined to become worse during a host of conditions, including social isolation, poverty, the hurried and high-pressure characteristics of present-day life, and a variety of other stressors. In this work, a comparative comparison was conducted of four machine learning models for classifying mental illness: Support Vector Machine, Naive Bayes, Logistic Regression, along with Decision Tree. We trained and tested these models using a dataset of mental health-related text. Accuracy, precision, recall, and F1 score were used to assess them. Evaluation of our tests demonstrated that SVM had the best and most constant accuracy and F1 score, outperforming both Logistic Regression, Decision Tree, and Naïve Bayes models. Advantages and disadvantages of each model are illustrated in this paper, along with recommendations on ways to enhance its functioning. As per our results, SVM is a promising approach for spotting mental illness in text data overall.