Modelos mixtos para datos composicionales: Una aplicacion con resultados electorales en Colombia
Fecha
2017Registro en:
Liscano, J. (2017). Modelos mixtos para datos composicionales: una aplicación con resultados electorales en Colombia. (Trabajo de pregrado). Universidad Santo Tomás. Bogotá, Colombia.
reponame:Repositorio Institucional Universidad Santo Tomás
instname:Universidad Santo Tomás
Autor
Liscano Fierro, Juan Manuel
Institución
Resumen
The present work consists in the application of certain tools developed for the analysis of the compositional
data. The purpose includes the revision of the theoretical aspects; the geometry of the simplex,
the log-ratio methodology and aspects related to null components, as well as the development of a practical
exercise taking into account the mentioned methodologies along with the statistical models, such as
Dirichlet regression, a multivariate linear model and nally the multivariate mixed model, which is the
main axis of the exercise. It illustrates the practical application of the theory making use of the available
information about the electoral processes carried out in Colombia and other variables that de ne the
economic and political situation of the country.
The results of the data analyzed under the adjustment of the mixed model respond in the best way to
the real values of the plebiscite, identifying how the variables worked in
uence the results of the voting.
Suggesting that departments with more social problems are more in favor of peace.