<![CDATA[This study examines the robustness of the K-Medoids algorithm to the presence of outliers in management data derived from Airbnb property listings, focusing on price and review rating attributes as key indicators in market segmentation and managerial decision-making. The increasing use of data from digital platforms is often accompanied by the emergence of extreme values that have the potential to disrupt the cluster structure and reduce the reliability of the analysis results. Therefore, a clustering method is needed that is able to maintain the stability and consistency of results even though the data contains outliers. This study uses a quantitative and experimental approach based on secondary data, where the implementation of the K-Medoids algorithm and visualization of clustering results are carried out using Python programming, while manual calculations based on Excel spreadsheets are used as a means of conceptual validation of the medoid selection process and distance measurement. The analysis is carried out by comparing the cluster results before and after the presence of outliers using the Manhattan distance metric and evaluating the total cost function at each iteration. The results show that although extreme values in the price attribute cause a wider spread in the data, the cluster structure and medoid positions remain relatively stable. The resulting clusters also retain clear managerial significance, grouping properties into economy, regular, and premium segments, thus remaining relevant to support pricing strategies, service differentiation, and reputation management.]]>
Copyrights © 2026
