Title: Statistical tools to deal with off-diagonal elements in square asymmetric matrices obtained from spatial data
Authors: Inmaculada Barranco-Chamorro - Universidad de Sevilla (Spain) [presenting]
Abstract: Change and confusion matrices are a quite common tool used in spatial statistics. They are usually obtained as a result from classification of images and fotointerpretations. From the statistical point of view, we have a square asymmetric contingency table where the rows and columns refer to the same set of objects, and their entries are non negative integer numbers. Quite often, standard statistical methods designed for contingency tables are unsuccessful to deal with this kind of matrices because of the strong effect of the diagonal values on the statistical summaries. So, we focus on matched-pairs statistical methods to deal with the off-diagonal elements of these matrices. Marginal homogeneity tests are considered where exact and approximate results are given. Practical applications to real spatial data are included.