Predicting academic performance is a crucial task in educational institutions, as it aids in identifying students who may require additional support or intervention. This research paper presents an innovative approach to academic performance prediction by combining K-Means clustering with an in-depth comparison of advanced machine learning algorithms. The study utilizes a comprehensive student dataset sourced from Kaggle, encompassing diverse attributes related to students’ backgrounds and academic history. The dataset is subjected to K-Means clustering, resulting in the identification of distinct student clusters. Subsequently, four prominent machine learning algorithms—K-Nearest Neighbors (KNN), Neural Network (NN), Random Forest (RF), and Support Vector Machine (SVM)—are rigorously evaluated for their predictive efficacy on academic performance. The experimental results showcase the utility of K-Means clustering in segmenting students into meaningful clusters based on shared characteristics. The comparative analysis of the machine learning algorithms reveals varying levels of accuracy, precision, recall, and F1-score in predicting academic performance across different clusters. The outcomes highlight the algorithm that exhibits superior performance in this specific context.