<![CDATA[The increasing use of short-form video platforms has created diverse patterns of digital media usage among communities. YouTube Shorts is not only used for entertainment but also for obtaining information, interacting with content, and creating digital content. These diverse behaviors make it necessary to identify user groups based on their similarities in using YouTube Shorts. This study aims to apply the K-Means algorithm to cluster community behavior in using YouTube Shorts in Ujung Bandar Village, Rantau Selatan District, Labuhanbatu Regency, North Sumatra. The research used a quantitative descriptive approach with primary data collected through a questionnaire. A total of 250 respondents were involved in the study. The behavioral data were represented using five variables: usage intensity, usage purpose, frequency of information utilization, interaction level, and content creation involvement. Data preprocessing included data cleaning, coding, tabulation, and the formation of variable values. The optimal number of clusters was determined using the Elbow Method, which indicated three clusters. The K-Means algorithm was subsequently implemented using Python through Google Colaboratory with Euclidean distance. The final clustering produced three groups consisting of 82 respondents in Cluster 1, 91 respondents in Cluster 2, and 77 respondents in Cluster 3. Based on the final centroid values, Cluster 1 represented medium usage behavior, Cluster 2 represented high usage behavior, and Cluster 3 represented low usage behavior. The clustering results were evaluated using the Davies-Bouldin Index and Silhouette Score, obtaining values of 1.1140 and 0.3426, respectively. The results provide a description of the diversity of YouTube Shorts usage behavior and can be used as supporting information for developing more targeted digital information and content strategies.]]>
