KOMPARASI VALIDASI K-MEANS DAN FUZZY C-MEANS DALAM MENENTUKAN CLUSTER TERBAIK MENGGUNAKAN SILHOUETTE INDEX DAN DAVIES BOULDIN INDEX
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Telkom Purwokerto
Author
TAUFIK, HIDAYAT
Subject
T Technology (General)
Datestamp
2021-04-26 03:18:17
Abstract :
ABSTRACT The size of Indonesia population must be balanced with health insurance that is evenly distributed in each region. The Indonesia government through the Ministry of Health annually collects population health data from the subdistrict to the provincial level. This is done to determine the ranking of healthy region and become an important reference for Regional Government (Pemda) to provide health assistance which is more targeted in formulating of Problematic Areas for Heavy/Special Health (DBKBK). In helping the government to formulate DBKBK, data mining is used to group the large data based on data similiarities. K-Means is one of the simplest and most widely used clustering algorithm and able to group large amounts of data with relatively fast and efficient time, while the Fuzzy CMeans is one of the fuzzy grouping models with membership levels differing between 0 and 1 so that data can bea member of a special cluster and each data has a membership distance for each cluster. This study aims to describe the best clusters in the regency/city grouping distribution based on the condition of the regional health profile wiith the best cluster algorithm. The data used in this study are district/city health data in 2017 with 550 data samples and five variables obtained from the Ministry of Health Data and Information Center (Pusdatin Kemkes). The result of this study indicate that K-Means algorithm provides the best results compared to Fuzzy C-Means for cluster K=3, the Davies Bouldin Index (DBI) validation value is 0.4571, and the Silhouette Index (SI) validation value is 0.7486. Keywords – Data Mining, Davies Bouldin Index, Clustering, Health, Silhouette Index
ABSTRACT The size of Indonesia population must be balanced with health insurance that is evenly distributed in each region. The Indonesia government through the Ministry of Health annually collects population health data from the subdistrict to the provincial level. This is done to determine the ranking of healthy region and become an important reference for Regional Government (Pemda) to provide health assistance which is more targeted in formulating of Problematic Areas for Heavy/Special Health (DBKBK). In helping the government to formulate DBKBK, data mining is used to group the large data based on data similiarities. K-Means is one of the simplest and most widely used clustering algorithm and able to group large amounts of data with relatively fast and efficient time, while the Fuzzy CMeans is one of the fuzzy grouping models with membership levels differing between 0 and 1 so that data can bea member of a special cluster and each data has a membership distance for each cluster. This study aims to describe the best clusters in the regency/city grouping distribution based on the condition of the regional health profile wiith the best cluster algorithm. The data used in this study are district/city health data in 2017 with 550 data samples and five variables obtained from the Ministry of Health Data and Information Center (Pusdatin Kemkes). The result of this study indicate that K-Means algorithm provides the best results compared to Fuzzy C-Means for cluster K=3, the Davies Bouldin Index (DBI) validation value is 0.4571, and the Silhouette Index (SI) validation value is 0.7486. Keywords – Data Mining, Davies Bouldin Index, Clustering, Health, Silhouette Index
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
