Generalized permutation polytopes and exploratory graphical methods for ranked data (Q1314462): Difference between revisions

Exploratory graphical methods for fully and partially ranked data are proposed. In fully ranked data, \(n\) items are preference ranked by a group of judges. In partially ranked data, the judges do not completely specify their ranking. The resulting set of frequencies is a function on the symmetric group of permutations if the data are fully ranked, and a function on a coset space of the symmetric group in the partially ranked case. Since neither the symmetric group nor its coset spaces have a natural linear ordering, traditional graphical methods such as histograms and bar graphs are inappropriate for displaying such data. For fully ranked data, the frequencies can be plotted naturally on the vertices of a permutation polytope (pp is the convex null of the \(n!\) permutation points in \(\mathbb{R}^ n\)) and the metrics Spearman's \(\rho\) and Kendall's \(\tau\) easily interpreted. For partially ranked data, a generalized permutation polytope (gpp) is defined similarly. The frequencies in this case can again be plotted in a natural way on the vertices of a gpp. Gpp's induce an extension of Kendall's \(\tau\) for the partially ranked case. Also, the fixed vector version of Spearman's \(\rho\) in this case is easily interpreted on gpp's. The problem of visualizing data plotted on polytopes in \(\mathbb{R}^ n\) is addressed by developing the theory needed to define all the faces (particularly the 3 and 4 dimensional faces) of any gpp. This is done by writing a pp as the intersection of a system of linear equations, and extending the results to gpp. The proposed graphical methods are then illustrated on five different data sets.