Open AccessOptimization of Unsupervised Learning in Machine Learning
Author(s) -
Hidayatus Sibyan,
Wildan Suharso,
Edi Suharto,
Melda Agnes Manuhutu,
Agus Perdana Windarto
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1783/1/012034
Subject(s) - cluster analysis , artificial intelligence , unsupervised learning , machine learning , computer science , quality (philosophy) , medoid , geography , philosophy , epistemology
The Ombudsman of the Republic of Indonesia (hereinafter referred to as the Ombudsman) is a state institution (independent) that has the authority to oversee the administration of public services. The purpose of this study is to analyze the completion of reports/complaints from the public by using unsupervised learning techniques in machine learning. The data source used is the statistical report/public complaints based on the classification of the reporter and how to submit it in each provincial regional office (simpel.ombudsman.go.id). The unsupervised learning techniques in machine learning that are used are clustering (k-medoids) and classification (C4.5) which are part of data mining. k-medoids is tasked with mapping community reports/complaints based on provincial regional offices. The results of the mapping will be classified to get the range of values from the existing mapping. The calculation process uses the help of RapidMiner software. The distribution labels used were 4 clusters namely the percentage of completion of the “very good” report (C1) of 9 provinces; percentage of “good” report completion (C2) of 10 provinces; percentage of completion of “lacking” reports (C3) of 11 provinces; percentage of “bad” report completion (C4) of 3 provinces. The Davies-Bouldin Index value on the map is 0.530 (optimal). The results of the mapping can be information in improving the quality of public services in the completion of the report including the provinces included in the C3 and C4 clusters with the percentage of report completion classification (C4 = 0 - 20.70% and 20.70% > C3 < 47.69%).