Implementation of the K-Means Cluster Analysis Algorithm to Determine Student Self-Regulated Learning (SRL) Groupings in Carrying Out e-Learning in Yogyakarta

This research implements the K-means cluster analysis algorithm to determine student SRL groupings when carrying out e-Learning. The research results reveal variations in students' SRL profiles during e-Learning, using two approaches in selecting the best model by optimizing the Silhouette value and the Elbow method. The first approach, using the Silhouette method, selected two clusters that had the highest Silhouette scores, namely 0.111 for cluster 1 and 0.232 for cluster 2. The number of members in Cluster 1 was 1645 students, while Cluster 2 had 2229 students. The results of this analysis show that there are differences in SRL characteristics between the two groups. The second approach, using the Elbow method with the best model is a model with five clusters. Although the Silhouette score value in this model is lower (0.090) than the model with two clusters, this approach can help to gain deeper insight into the characteristics that may exist in the data. These clusters have different SRL characteristics. The results of student cluster analysis using the K-means algorithm provide a deeper understanding of student characteristics in organizing their learning. This research provides a deeper look at how e-Learning influences students' SRL and how the use of cluster analysis can support this understanding.