Ensemble ROCK Methods and Ensemble SWFM Methods for Clustering of Cross Citrus Accessions Based on Mixed Numerical and Categorical Dataset

3rd International Seminar On Sciences "Sciences On Precision And Sustainable Agriculture"

Alvionita^1 ; Sutikno^1 ; Suharsono, A.^1

Department of Statistics, Sepuluh Nopember Institute of Technology, Indonesia^1

关键词: And filters;    Categorical data;    Clustering Ensemble;    Clustering methods;    Ensemble methods;    Multivariate analysis methods;    Numerical data;    Standard deviation;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/58/1/012029/pdf
DOI  :  10.1088/1755-1315/58/1/012029

来源: IOP

【 摘 要 】

Cluster analysis is a technique in multivariate analysis methods that reduces (classifying) data. This analysis has the main purpose to classify the objects of observation into groups based on characteristics. In the process, a cluster analysis is not only used for numerical data or categorical data but also developed for mixed data. There are several methods in analyzing the mixed data as ensemble methods and methods Similarity Weight and Filter Methods (SWFM). There is a lot of research on these methods, but the study did not compare the performance given by both of these methods. Therefore, this paper will be compared the performance between the clustering ensemble ROCK methods and ensemble SWFM methods. These methods will be used in clustering cross citrus accessions based on the characteristics of fruit and leaves that involve variables that are a mixture of numerical and categorical. Clustering methods with the best performance determined by looking at the ratio of standard deviation values within groups (SW) with a standard deviation between groups (SB). Methods with the best performance has the smallest ratio. From the result, we get that the performance of ensemble ROCK methods is better than ensemble SWFM methods.