Grouping of Flood Victim Data Based on Damage Rate Using K-Means Algorithm Case Study: Binjai City Social Service

Abstract

This study aims to classify flood disaster victim data in Binjai City based on three main variables: sub-district or location, damage level, and type of aid received. The data were obtained from the Binjai City Social Service in 2024 and processed using the K-Means Clustering method with the Matlab R2014b application. The stages include data transformation, determining the number of clusters, selecting initial centroids, calculating Euclidean distance, and evaluating the results. Tests were conducted with configurations of 2, 3, 4, and 5 clusters. In the 2-cluster configuration, the distinction was observed between areas with low damage and limited to moderate aid, and areas with medium damage and more extensive aid. In the 3-cluster configuration, the second test produced the most optimal cluster in Kartini Sub-district with light damage and limited food aid. In the 4-cluster configuration, the most compact cluster was found in Setia Sub-district with medium damage and aid in the form of food and blankets. In the 5-cluster configuration, the most specific result was obtained in Rambung Barat Sub-district with medium damage and aid in the form of food and blankets. These findings indicate that the 5-cluster configuration provides more detailed and targeted classification, serving as a strategic reference for aid distribution.