ARCHIVES

Social media platforms have revolutionized communication, offering users a vast variety of opportunities to connect and share ideas. However, this freedom has also led to a rise in cyberbullying, which significantly impacts mental health and well-being. Cyberbullying often involves complex language, sarcasm, slang and subtle threats, making it difficult for automated systems to accurately identify. This research presents a supervised predictive analytic method for detecting cyber bullying on social media using Logistic Regression. The primary objective is to design an efficient system that can identify and classify cyber bullying incidents early; helping to prevent their escalation. Logistic Regression was employed as the core algorithm to predict the presence of cyber bullying. The model demonstrated proving the reliability of Logistic Regression in text classification tasks. Furthermore, additional analysis was performed to assess how various feature engineering techniques influence model performance. The research emphasizes the significance of incorporating diverse linguistic and contextual cues to enhance the accuracy of cyber bullying detection. In conclusion this project contributes to the proactive identification of cyber bullying by offering a scalable and effective solution using Logistic Regression, thus supporting safer online interactions on social media platforms.