Mental health interpretation from text is essential for early identification and successful intervention. Through sentiment analysis of text data, including social media and online communication, useful information can be obtained about people’s mental states, making it simpler to detect potential problems before they are aggravated. This approach enhances the ability to monitor and assess mental health status and enables the development of preemptive intervention plans, ultimately providing improved outcomes for mentally ill patients. The study provides a model that involves the integration of a Support Vector Classifier (SVC) and a Term Frequency-Inverse Document Frequency (TF-IDF) vectorizer for the implementation of sentiment analysis on tweets. The model demonstrated remarkable effectiveness in the paradigm of binary classification, when Depression and Suicidal classes were combined into a single class with an accuracy rate of 95.33% and a macro average F1-score of 0.95. In the paradigm of multiclass classification, which divided into the different classes of Depression, Normal, and Suicidal, the model delivered accuracy of 81.40% and a macro average F1-score of 0.79. Although multiclass performance is encouraging, the latter is less impressive compared to binary classification performance. The results demonstrate the efficacy of the SVC with TF-IDF in sentiment classification and improve the model’s ability to interpret and respond to mental health from text data.