Teste de associação espacial entre dois tipos de polígonos

• Ano de defesa: 2019
• Autor(a) principal: Lucas da Cunha Godoy
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; Programa de Pós-Graduação em Estatística; UFMG
• Palavras-chave em Português: Padrões de polígonos espaciais; Associação espacial; Testes de Monte Carlo Condicionais; Toroidal Shift; Estatística - Teses; Polígonos; Método de Monte Carlo; Análise espacial (Estatística)
• Link de acesso: http://hdl.handle.net/1843/33511

It is common in spatial statistics the interest in testing hypotheses related to the spatial association between two spatial patterns. Often the spatial entities that best represent the data are polygons. However, the complexity and lack of techniques developed in this field forces researchers to simplify such polygons by points, as their centroids. In this work, techniques were developed to test hypotheses related to polygons patterns. These techniques take advantage of all the information present in these spatial entities. The methodology developed here is centered on the toroidal shift. The use of this method brings challenges that caused other authors to disregard it, even though it is a better null model alternative. A simulation study was also carried out to prove the efficacy of the developed methodology and to compare different test statistics. Finally, the techniques were applied to real data regarding different species of short grams in the state of Colorado.