Main Article Content
Abstract
Clustering data through hierarchical approach could be performed by Agglomerative Nesting (AGNES) Method and Divisive Analysis (DIANA) Method. The objective of this research is to compare both the methods based on Euclid and Manhattan distance measurements. Of this research the clustering procedures of agglomerative method are conducted by exploring all techniques including single linkage, complete linkage, average linkage, and Ward. The data used are the National Socio-Economic Survey (SUSENAS) data which are selected specifically for the percentage of over 5 year old residents in each province, for both living in urban or rural, who access the internet in the last 3 months in 2017 but classified according purpose of accessing. By applying Mean Square Error (MSE) for 2 and 3 clusters, it can be concluded that the single linkage technique is the best performance of clustering procedure for both Euclidean and Manhattan distances.
Keywords
Article Details
References
- Timm, N.H., Applied Multivariate Analysis, New York, Springer-Verlag. Inc., 2002.
- Everitt, B, Cluster Analysis, Third Edition, New York, Halsted Press, 1993
- Gan, G., Ma, C., and Wu, J., Data Clustering: Theory, Algorithms, and Applications, ASA SIAM series on Statistics and Applied Probability, Philadelphia, 2007.
- Wishart, D., “K-Means Clustering with Outliers Detection, Mixed Variables and Missing Values”, Explanatory Data Analysis in Empirical Research, pages 216 – 226, New York, Springer, 2002
References
Timm, N.H., Applied Multivariate Analysis, New York, Springer-Verlag. Inc., 2002.
Everitt, B, Cluster Analysis, Third Edition, New York, Halsted Press, 1993
Gan, G., Ma, C., and Wu, J., Data Clustering: Theory, Algorithms, and Applications, ASA SIAM series on Statistics and Applied Probability, Philadelphia, 2007.
Wishart, D., “K-Means Clustering with Outliers Detection, Mixed Variables and Missing Values”, Explanatory Data Analysis in Empirical Research, pages 216 – 226, New York, Springer, 2002