GROUPING OF PROVINCES IN INDONESIA BASED ON MATERNAL HEALTH INDICATORS USING VARIABLE WEIGHTING K-MEANS AND FUZZY C-MEANS

Maternal Mortality Rate (MMR) is an indicator used to assess maternal health as well as the health status of a country. MMR is a target that must be achieved by Indonesian Government in Sustainable Development Goals (SDGs) in 2030. The Government of Indonesia has made various efforts to reduce MMR. This study aims to determine the distribution pattern of indicators for improving maternal health by grouping provinces in Indonesia based on the characteristics of maternal health indicators. The variables used are indicators that affect maternal mortality, namely K4 coverage (x1), Td2+ immunization coverage (x2), maternity assisted by health workers in health facilities coverage (x3), post-partum check up coverage (x4), Puskesmas implementing pregnant classes (x5), Puskesmas implementing P4K (x6), participant of KB coverage (x7) in Indonesia in 2017. The grouping methods are Variable Weighting K-Means (VWKM) and Fuzzy C-Means (FCM). The selection of the best grouping results uses the Internal Cluster Dispersion Rate (icdrate). Based on the analysis results, the best grouping is generated by the FCM method. The icdrate value generated by FCM is 0.325 while the icdrate value generated by VWKM is 0.552. FCM produces five groups which can be categorized as groups with maternal health indicator characteristics with very low, low, medium, high, and very high scores. Provinces in a group tend to be geographically close. East Java and Bali are provinces included in the indicator group of very high maternal health. Papua and West Papua fall into the group for maternal health which is very low.

Co-regularized weighting multiview clustering

This paper deals with clustering for multiview data. Multiview clustering has been a research hot spot in many domains or applications, such as information retrieval, biology, chemistry, and marketing. Exploring information from multiple views, one can hope to find a clustering that is more accurate than the ones obtained using the individual views. The aim is to search for clustering patterns that perform a consensus between the patterns from different views. Inspired by variable weighting and co-regularized strategy, this paper studies co-regularized weighting multiview clustering algorithms. Two co-regularized weighting multiview clustering algorithms are proposed from two aspects: pairwise co-regularization and centroid-based co-regularization. Experimental results obtained both on synthetic and real datasets show that the proposed algorithms outperform the main existing multiview clustering algorithms.