Claim Missing Document
Check
Articles

Found 2 Documents
Search
Journal : Jurnal Mantik

Aplication of Validity Index in K Means and Fuzzy C Means Jontinus Manullang; Pahala Sirait; Andri Andri
Jurnal Mantik Vol. 4 No. 2 (2020): Augustus: Manajemen, Teknologi Informatika dan Komunikasi (Mantik)
Publisher : Institute of Computer Science (IOCS)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35335/mantik.Vol4.2020.958.pp1430-1438

Abstract

K-Means and Fuzzy C-Means Clustering is a method of analyzing data that performs the modeling process without supervision (without supervision) and is a method that groups data by partitioning the system. Clusters Clusters and Fuzzy C-Means will produce different clusters in the same dataset, cluster validity index is a method that can be used to improve the results of clustering generated by the clustering method. This study will use the cluster validity index on the kmeans clustering algorithm and Fuzzy C-Means by calculating the index of validity of each kmeans clustering result with k = 2, ..., kmax (k max determined at the beginning) and the results from Fuzzy C-Means with c = 2, ...., cmax (c max is specified at the beginning). By using the cluster validity index, the most optimal cluster is obtained in the second cluster with the Dbi value = 0.45 in the mean K and the second cluster with the Dbi value = 0.5 in the Fuzzy C Mean, and the results of the clustering are consistent.
Credit Card Risk Classification Using K-Nearest Neighbor Weighted Algorithm Based on Forward Selection Sartika Dewi Purba; Pahala Sirait; Arwin Arwin
Jurnal Mantik Vol. 4 No. 3 (2020): November: Manajemen, Teknologi Informatika dan Komunikasi (Mantik)
Publisher : Institute of Computer Science (IOCS)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35335/mantik.Vol4.2020.960.pp1551-1559

Abstract

One form of credit card risk is non-performing credit cards, which describe a situation where loan repayment approval on credit cards runs the risk of failure. In the classification technique there are several algorithms that can be used, one algorithm that is often used is Weighted k-nearest neighbor (WKNN). This study aims to improve the performance of the Weighted k-nearest neighbor (WKNN) algorithm by applying the forward selection feature that is used to select each unused feature when starting a feature iteration, the results of the study show that by adding forward performance selection of the Weighted k-nearest algorithm neighbor (WKNN) get a better value that is 86.4%, compared to using the Weighted k-nearest neighbor (WKNN) algorithm without a forward selection that is equal to 60.1%.