With the proliferation of Internet usage in contemporary society, a vast amount of data has been generated. However, alongside its benefits, the cyber world presents its own array of challenges, among which cyberbullying stands out as a significant issue. Cyberbullying, constituting an online form of crime, encompasses various criminal activities facilitated through the Internet, computers, mobile phones, and other electronic devices. Previous research in detecting cyberbullying has encountered limitations, including data unavailability, hidden identities of perpetrators, and victims' privacy concerns. To address these constraints, this paper proposes an effective text mining approach utilizing machine learning algorithms to actively identify bullying texts. Unlike prior studies focusing solely on textual features, this study incorporates three distinct types of features: textual, behavioral, and demographic. Textual features encompass intimidating language indicative of potential cyberbullying outcomes. Meanwhile, behavioral features are derived from observed online actions, and demographic features include age, gender, and location information extracted from the dataset. By integrating these multifaceted features, the proposed method aims to enhance the accuracy and robustness of cyberbullying detection on social media platforms.