Application of Cluster Analysis of Self Organizing Map (SOM) Method in the Community Literacy Development Index in Indonesia

Abstract

Self Organizing Map (SOM) is a method with a form of unsupervised learning, with Artificial Neural Network (ANN) training techniques that use a winner takes all basis, where only the neuron that is the winner will be updated. This study applies the cluster analysis of the SOM method in grouping provinces in Indonesia based on the characteristics of the Community Literacy Development Index (IPLM). The selection of the best cluster is based on internal validation i.e. connectivity, index Dunn and Silhouette. Based on the cluster validation results, 3 clusters were obtained that group provinces based on IPLM characteristics. of the 7 (seven) elements that make up the IPLM, 2 of them, namely energy and community visits, are shown in cluster 1. 5 other elements such as libraries, collections, SNP libraries, community involvement and library members are shown in cluster 3. Meanwhile, cluster 2 does not show significant IPLM-forming elements.