Comparison of Cluster Ensemble and Two Step Cluster Methods on Clustering with Mixed Type Data
Fera Hermawati, Budi Susetyo, Agus Mohamad Soleh · Zenodo (CERN European Organization for Nuclear Research) · 2018
Health development is supported by the availability of adequate health facilities and personnel. To facilitate the government in determining the policies taken, it is necessary to group the region to know which areas that need improvement in health facilities and personnel. Cluster analysis is used to group objects based on certain characteristic similarities. Cluster analysis is generally applied to objects with numerical data types. Health facility and health personnel data have categorical and numerical types or also called mixed data type, so it is necessary to use clustering for mixed types data. This study aims to compare cluster ensemble method and two step cluster method in clustering mixed type data. The comparative criterion used is the ratio between diversity within cluster (S_w) and the diversity between cluster (S_b). Smaller ratio values indicate a better method. The research results showed that cluster ensemble method is a better method than the two step cluster method in clustering mixed type data.