Principal Component Analysis (PCA) is a popular method used for reduction of large scale data sets in hydrological applications. Typically, PCA scores are applied to hierarchical cluster analysis to redefine region. However, the choice of cumulative percentage of variance for PCA scores and identifying the best number of clusters can be difficult. In this paper, we investigate the effect of determining the number of clusters by comparing (i) standardized and unstandardized PCA scores on different cumulative percentages of variance (ii) to determine number of clusters using Calinski and Harabasz Index. We have found that Calinski and Harabasz Index is most appropriate to determine the best number of clusters and that standardized PCA scores within the range of 65% to 70% cumulative percentage of variance give the most reasonable number of clusters.
CITATION STYLE
Shaharudin, S. M., & Ahmad, N. (2017). Choice of cumulative percentage in principal component analysis for regionalization of peninsular Malaysia based on the rainfall amount. In Communications in Computer and Information Science (Vol. 752, pp. 216–224). Springer Verlag. https://doi.org/10.1007/978-981-10-6502-6_19
Mendeley helps you to discover research relevant for your work.