<![CDATA[This study aims to analyze the effect of distance metrics on the K-Medoids algorithm in clustering teacher performance based on pedagogical, personality, social, and professional indicators. This study was conducted because previous research on teacher performance clustering has generally focused more on the use of clustering algorithms without comparatively evaluating the effect of distance metric selection on the quality of the resulting clusters. The research data were obtained from assessor evaluations of teacher performance at SD Islam Al-Irsyad involving 16 teacher data records, which were first normalized to standardize the scale among attributes before the clustering process was carried out. Clustering was performed using the K-Medoids algorithm with Euclidean, Manhattan, and Cosine distance methods, as well as variations in the number of clusters (K) from 2 to 5 using RapidMiner. The clustering results were evaluated using the Silhouette Coefficient method calculated with Python to determine the cluster quality and the optimal number of clusters. The results showed that the highest Silhouette Coefficient value of 0.290 was obtained at K=4 using the Euclidean and Manhattan methods. Based on these results, the Euclidean method was selected as the best method because it is more widely used and capable of representing the distance between data points more effectively. The clustering results were then analyzed by assigning labels to each cluster based on the average value of the indicators to facilitate interpretation. Thus, this study demonstrates that the selection of distance metrics affects the quality of clustering results in teacher performance grouping.]]>
