Estudo comparativo entre as funções de ligação Logit e Probit : estimando parâmetros

• Autor(a) principal: Jesus, Matheus Francisco Nascimento de
• Data de Publicação: 2015
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: por
• Título da fonte: Repositório Institucional da UFS
• Texto Completo: http://ri.ufs.br/jspui/handle/riufs/8337
• topic: Estatística; Ensino de estatística; Modelos lineares generalizados; Software R.; Generalized linear models; Software R.; CIENCIAS EXATAS E DA TERRA::PROBABILIDADE E ESTATISTICA::ESTATISTICA::ANALISE DE DADOS

This study has a good introduction to the family of exponential distributions which is an indispensable subject when it comes to Generalized Linear Models and also good content on the latter. The logit and probit link functions in order to evaluate the performance of the same in the estimation of parameters with changed data were used. Three different fixed values for the parameter beta, which is a probability of success, then simulations were performed in 1000 (estimates) for each combination of parameters and connection functions. We used the mean squared error and the percentage of children squared errors than 0.02 to be used for comparison. Came to the conclusion that the probit link function was indicated to estimate the parameters next to zero and that the logit link function was the most suitable for the estimation of parameters next to one. The whole procedure was performed using the free software R.