CASE BASE REASONING UNTUK MENDIAGNOSIS PENYAKIT HIPERTENSI MENGGUNAKAN METODE INDEXING DENSITY BASED SPATIAL CLUSTERING APPLICATION WITH NOISE (DBSCAN)

Santoso, Herdiesel and Syafrianto, Andri (2019) CASE BASE REASONING UNTUK MENDIAGNOSIS PENYAKIT HIPERTENSI MENGGUNAKAN METODE INDEXING DENSITY BASED SPATIAL CLUSTERING APPLICATION WITH NOISE (DBSCAN). Ethos : Jurnal Penelitian dan Pengabdian kepada Masyarakat, 7 (1). pp. 88-100. ISSN 2502-065x

Text Naskah Jurnal_CBR Untuk Mendiagnosis Penyakit Hipertensi_compressed.pdf Download (458kB)

Abstract

Hypertension is one of the health problems priority in the world because of the increasing of life expectancy and an unhealthy lifestyle. Many people with hypertension are unreachable and undiagnosed by a health worker and they do not do treatment according to the health recommendation. The Case-Based Reasoning (CBR) Method can be applied to solve the new cases in diagnosed hypertension using the answer or experience from the old case by comparing the new case and the old case. In order to do not use all the basic case data for finding a similar case, it makes an indexing process is needed. The DBSCAN algorithm implementation as indexing method is expected to improve the time and memory efficiency in CBR, especially during the retrieval process. The result of the CBR test with the cluster-indexing has a better accuracy and time process than the non-indexing CBR. The minimum parameter points and epsilon that has been chosen for clustering on hypertension data case is the combination of epsilon score 9 and minimum points score 3 with the silhoutte coefficient score 0.240 and average cluster time 0.541 seconds. The Minkowski Distance method has better accuracy than the Euclidean Distance method because by the threshold score ≥ 0.9 the CBR system with the Minkowski distance method is able to diagnose the disease with 100 % accuracy and the average best retrieval time, it is 0.0586 second

Actions (login required)

View Item