Welfare Mapping of Region in Indonesia based on Health and Nutrition Indicators using Self Organizing Mapping (SOM) Method

Abstract

Health and nutrition are important indicators of public welfare. Availability map of the region on those indicators is necessary for constructing an effective plan of welfare equal distribution in health and nutrition. This study aims to create a map of the welfare index on health and nutrition indicators over Indonesian provinces. This research employs a soft computing approach specifically a self-organizing mapping (SOM). SOM is a method based on the neural network concept. The data are drawn from the Statistics Indonesia. Sixteen variables are considered as factors to create the map. They could be grouped into six factors namely illness, mortality, life expectancy, nutritional conditions, breastfeeding and immunization, clean water facilities and latrines. A Davies Bouldin Index (DBI) is used to determine the optimal number of clusters. The best mapping is obtained by five clusters as corresponding to the least DBI value. Thirty-four provinces in Indonesia are mapped into five clusters. The characteristics of each cluster are described by six factors of health and nutrition indicators.