Perbandingan K-Means, K-Medoids dan GMM untuk Pengelompokan Kualitas Lulusan Guna Penjaminan Mutu Perguruan Tinggi

The student graduation rate is one of the parameters of the evaluation of the university learning process. Where the study period must be taken according to the required time range. In addition, to measure the graduation rate is seen from 3 things, namely the study period, the Cumulative Achievement Index (GPA) and the Excess Semester Credit Unit (SKS). To obtain the results of a good graduate quality analysis results, the best data processing method is needed. The purpose of this study is to examine the performance of the K Means, K Medoids, and Gaussian Mixture Models (GMM) methods with the cluster performance evaluation method in order to obtain optimal k to classify the quality of graduates for university quality assurance. The research methods studied are K-Means, K-Medoids, and GMM. The research stage consists of the data collection stage, data preparation, modeling with K-Means, K-Medoids, and GMM, Silhouette Coefficient cluster evaluation. The cluster evaluation with the silhouette score of the K-Means method with Z-Score normalization was obtained better than other methods with a value of 0.3487 in k-2. The study program that is a priority for coaching by the quality assurance unit is in cluster two, namely Farmasi, Mining Engineering. This research is expected to be a recommendation for methods on the problem of grouping.