ABSTRACT
This chapter compares and contrasts the relative strengths and weaknesses of different ways of viewing a Euclidean data matrix. This includes heat-map views where matrix entries are coded as colors, as well as treating each of the sets of matrix rows and columns as bundles of curves, to which Functional Data Analysis methods can be applied. A less standard combined view showing all of the above in a single plot, which is then used to display multiple modes of variation is proposed. Data centering of both rows and columns of the data matrix is explored, and appropriate OODA based terminology is developed to keep these often slippery concepts straight in interdisciplinary discussions. The scores scatterplot matrix view, for understanding relationships between data objects, is also considered in detail. That motivates discussion of a number of insightful alternatives to the standard PCA directions for visual scatterplot exploration of data.
