Grouping the districts in Indonesia based on value of science subjects of National Exam using K-Means clustering method

Abstract

Implementation of the National Examination (NE) for senior high schools for each district in Indonesia is done as one way to mapping the success of education in each the district. Mapping can be done by grouping districts based on attribute values obtained on the implementation of the NE, which in this study using the K-means clustering method. The attribute of NE values used are the values for the science majors, which consist of 6 subjects, where from 514 districts only 510 districts own it. The purpose of the study in addition to grouping districts, will also predict the group of 4 districts that have no NE values. Prediction is done by entering the value of the district Human Development Index (HDI) that has no UN value into 95% confidence interval of HDI from groups formed from k-means clustering. Once known where the district is in a group based on the HDI group then the value of its UN will follow the confidence interval of the UN value of the group. The results of this study indicate that the number of clusters (k) selected to classify districts in Indonesia as many as 5 clusters. Group prediction results show that one district belongs to the V cluster, while the other three districts are predicted to be members of the I cluster.