Comparative Analysis of Clustering Algorithms for NOUL Entrance Examination Data

The main objective of this study is to compare machine learning techniques for clustering, specifically using the K-Means and DBSCAN algorithms. The paper involved building, testing, and evaluating clustering models on real-world data. The dataset, collected from the National University Entrance examination (2014-2015), contains 15,602 samples. The Elbow method was applied to determine the optimal k value for K-Means, while the optimal epsilon value was used for DBSCAN. Python served as the development environment. The results showed that the K-Means model produced three clusters with a silhouette coefficient of 0.457, while DBSCAN achieved a higher silhouette score of 0.484. Therefore, DBSCAN outperformed K-Means in clustering performance.