Comparison of Agglomerative Hierarchical and K-Means in Grouping Provinces Based on Maternal Health Services

Abstract

D uring the Covid-19 period, there were barriers to access for pregnant women to health services that could interfere with maternal health. Therefore, it is necessary to know  the achievement of maternal health service coverage in Indonesia during the Covid-19 period in 2020, especially at the provincial level so that it can help the government to determine regional priorities for the fulfillment of more adequate maternal health services. Determination of provincial priorities for the fulfillment of maternal health services can be achieved by grouping the regions according to the characteristics of maternal health services in the local province. Cluster analysis is able to group objects in the form of provinces into one cluster. The clustering methods that will be used are agglomerative hierarchical clustering and k-means clustering. The results of the clustering of the two methods will be compared with internal validation in the form of dunn index, connectivity index, ang silhouette index. The best clustering resuls are obtained by using agglomerative hierarchical clustering alghoritm using the complete linkage similarity function with the resulting five clusters. The results of the identification of cluster characteristics show that cluster 1 with 14 members is categorized as provinces with good coverage of maternal services. Cluster 2 which consists of 15 provinces is categorized as best coverage. Cluster 3 which member are NTT and Maluku is categorized as bad. Cluster 4 which member is East Kalimantan is categorized as sufficient coverage. Meanwhile cluster 5 which member are Papua and West Papua is still on concern because its categorized as worst coverage