Abstract
In this paper, we propose a derivative of interval-valued functional data, and apply it to cluster analysis. Symbolic Data Analysis (SDA) proposed by Diday is a new approach for analyzing various data which are represented by symbolic concept including numerical data, interval-valued data and modal-valued data. Functional Data Analysis (FDA) developed by Ramsay is an approach for analyzing functional data. In FDA, we can make good use of "smoothness" in datasets. Toyoda et al. (2009) proposed a clustering method for interval-valued functional data which is an extension of functional data. We discuss a derivative of interval-valued functional data with a pair of functions, an upper function and a lower function. We apply it to cluster analysis using a dissimilarity based on Hausdorff distance.
