The growing field of educational data mining seeks to analyse educational data in order to develop models for improving education and the effectiveness of educational institutions. Educational data mining is utilised to develop novel approaches for extracting information from educational databases, enabling improved decision-making within the educational system. The main objective of this research paper is to investigate recent advancements in data mining techniques within the field of educational research, while also analysing the methodologies employed by previous researchers in this area. The predictive capabilities of various machine learning algorithms, namely Logistic Regression, Gaussian Naive Bayes, Support Vector Machine, Random Forest, K-Nearest Neighbour, and XGBoost Classifier, were evaluated and compared for their effectiveness in determining students' academic performance. The utilisation of Random Forest and XGBoost classifiers in analysing scholastic, behavioural, and additional student features has demonstrated superior accuracy compared to other algorithms. The training and testing of these classification models achieved an impressive accuracy rate of approximately (96.46% & 87.50%) and (95.05% & 84.38%), respectively. Employing this technique can provide educators with valuable insights into students' motivations and behaviours, ultimately leading to more effective instruction and reduced student failure rates. Students' achievements significantly influence the delivery of education.