CLUSTERING MODEL OF LECTURERS PERFORMA IN PUBLICATION USING K-MEANS FOR DECISION SUPPORT DATA

This study aims to produce a clustering model using the K-Means algorithm built to map the performance of lecturers' publications. The method used was research and development which includes the stages of data collecting, data preprocessing, clustering process and cluster analysis. The input data consists of 87 with 8 attributes, namely number of articles in Sinta indexed journal, number of articles in Scopus indexed journal, number of citation in Scopus, H-index in Scopus, number articles in Google Scholar indexed journal, number of citation in Googe Scholar, H-index in Google Scholar and H-index10 in Google Scholar. The K-Means algorithm was used with 3 clusters and 100 epoches. The results of clustering were distributed in 3 clusters, namely cluster 1 with 17 members, cluster 2 with 32 members and cluster 3 with 38 members. The results of the cluster analysis with the identification of cluster members and the value of the cluster centroid indicated that cluster 1 is lecturers with relatively high publication performance performance and cluster 3 shows relatively low publication performance performance. The data from clustering results can be used for decision support model input data for broader lecturer performance performance.