Education Data Mining (EDM) has emerged as a critical field for uncovering patterns, trends, and actionable insights from educational datasets to improve learning outcomes and teaching methodologies. This survey synthesizes research conducted between 2020 and 2025, reviewing 30 published articles that explore the application of various machine learning (ML) techniques in EDM. The findings highlight the increasing integration of ML approaches in analyzing student behavior, predicting performance, personalizing learning paths, and optimizing institutional decision-making processes. The surveyed research reveals that ML-driven EDM is not only enhancing educational processes but also addressing critical challenges like dropout rates, assessment quality, and resource allocation. Furthermore, hybrid models combining multiple ML techniques are showing promise in increasing accuracy and robustness in predictive and prescriptive analytics. However, significant challenges remain. The articles underscore the importance of addressing ethical considerations, ensuring data privacy, and managing the biases inherent in educational datasets. Additionally, scalability and the adoption of EDM solutions in diverse educational settings require further exploration. This survey underscores the transformative potential of machine learning in education, fostering more inclusive and effective learning environments. It serves as a foundation for continued research and practical applications, aiming to bridge the gap between data-driven insights and real-world educational improvements. The findings emphasize the need for collaborative efforts among educators, researchers, and technologists to harness the full potential of EDM.