Implementation of Spatial Constraints in Clustering Algorithms: A Study on Basic Infant Immunization in Lamongan District During the COVID-19 Pandemic
Mohammad Dian Purnama, Moh Jainur Miko Wahyudi · International Journal of Computer Science and Humanitarian AI · 2024
Algorithms for clustering data are important for data analysis, especially when finding patterns in big datasets. Nevertheless, the spatial limitations that are important in real-world contexts are generally ignored by the classic clustering approaches. Spatial variables have become more significant in the health industry, particularly during the COVID-19 pandemic, in terms of assessing population requirements and the allocation of healthcare resources. The purpose of this work is to investigate the use of spatial restrictions in clustering algorithms and to apply this method to COVID-19 immunization data from Lamongan District. The data analysis includes the 27 subdistricts of Lamongan District for the year 2021. Based on the peak of the COVID-19 pandemic, which had a major effect on baby immunization coverage, 2021 was chosen. The four basic baby immunization coverages—DPT-HB-Hib3, Polio 4, Measles, and BCG—are the variables that are used. Two methods are used: a neighborhood-like hierarchical clustering algorithm and spatial limitations. Distance-based spatial weights are better than proximity-based spatial weights when it comes to spatial constraints. This method is employed because an infant's coverage of all essential vaccines may be impacted by the spatial structure. We discovered that the fundamental baby immunization variable formed five clusters. It was discovered that cluster five had the highest immunization coverage among all the clusters. The three sub-districts that make up this cluster are Mantup, Kembangbahu, Tikung, Sarirejo, Deket, Glagah, and Karangbinangun.