Application of K-Medoids Cluster Result with Particle Swarm Optimization (PSO) in Toddler Measles Immunization Cases

Abstract

The objective of the research was to analyze the clustering method by optimizing Particle Swarm Optimization (PSO) in the case of measles immunization for children under the age of 5. The research data source used is the North Sumatra Province Central Bureau of Statistics (https://sumut.bps.go.id/). The data used in 2019 included 33 records with variables BCG, DPT-HB3/DPT-HB/Hib3, CAMPAK (MEASLES + RUBELLA), POLIO 4 and HEPATITIS B. The methods used were k-medoids and PSOs. This method is used to determine the value of the Davies Bouldin Index (DBI) before the cluster value is determined (k). The resulting k values were compared to k-medoids without PSO. The best results of k-medoid and PSO will be tested by classification to see the accuracy value of the cluster formed. The results showed that the optimization of k-medoids and PSO was better with the number of clusters (k = 5) than 0.078. (DBI). The results also show that the accuracy of the cluster formed is 95% with a correlation of 0.98.