The goal of this study is to use machine learning approaches to detect depression indications in social media articles. Data gathering, pre-processing, feature extraction, model training, and performance evaluation are all aspects of the research. The collection consists of social media messages classified into three categories: not depressed, somewhat depressed, and severely depressed. The study contributes to the growing field of social media data-driven mental health analysis by stressing the use of feature extraction algorithms for obtaining relevant information from text data. The use of social media communications to detect depression has the potential to increase early intervention and help for people at risk. Several feature extraction approaches, such as TF-IDF, Count Vectorizer, and Hashing Vectorizer, are used to quantitatively represent textual data. These features are used to train and evaluate a wide range of machine learning models, including Logistic Regression, Random Forest, Decision Tree, Gaussian Naive Bayes, and Multinomial Naive Bayes. To assess the performance of the models, metrics such as accuracy, precision, recall, F1 score, and the confusion matrix are utilized. The Random Forest model with Count Vectorizer had the greatest accuracy on the development dataset, coming in at 92.99 percent. And with a macro F1-score of 0.362, we came in 19th position in the shared task. The findings show that machine learning is effective in detecting depression markers in social media articles.