Penerapan PCA dan Algoritma Clustering untuk Analisis Mutu Perguruan Tinggi di LLDIKTI Wilayah IV

Abstract

The Internal Quality Assurance System (SPMI) is a guideline used by universities to assess the quality of performance and implementation of higher education internally. SPMI is very important to be considered by universities in order to compete positively with other universities, both at home and abroad, as well as to improve the management and implementation of higher education in the institution. In this study, three machine learning algorithms are applied, namely K- Means, Mean Shift, and DBSCAN, to cluster SPMI data. The methods used include Principal Component Analysis (PCA) to reduce data complexity without losing important information, and three clustering algorithms to group universities based on similarity of quality indicators. The K-Means algorithm clusters data based on distance to the nearest centroid, Mean Shift identifies clusters based on data density, and DBSCAN clusters data based on density and is able to handle outliers and irregularly shaped clusters. The results show that Mean Shift produces the best cluster with Silhouette Score 0.566, Davies- Bouldin Index 0.648, and Calinski-Harabasz Index 971.07. The K-Means algorithm provides quite good results with Silhouette Score 0.466, Davies-Bouldin Index 0.757, and Calinski-Harabasz Index 757.06. Meanwhile, DBSCAN has lower performance with Silhouette Score 0.216, Davies-Bouldin Index 1.045, and Calinski-Harabasz Index 105.67. This research provides the results of identifying universities that need special attention and helps in strategic planning for quality improvement so that they can carry out guidance more effectively and contribute to the development of a quality assurance system for higher education in Indonesia.