La estimación de proporciones mediante técnicas Bayesianas

The estimation procedures based on Bayes' Theorem are still an unusual option in many of the environments of classic parametric inference. The aim of this paper is to show an effective scheme for the use of Bayesian estimation of unknown parameters. We have opted to focus on the estimation of parameters under the assumption of a binomial model, so that it can be followed by all those situations that meet the aforementioned probabilistic model. This approximation was studied in comparison with the classic parametric approximation, both in its point version and by means of interval estimation. On a study, by simulating samples of several sizes, we obtained empirical evidence regarding the advantage of the Bayesian procedure.ResumenLos procedimientos de estimación basados en el teorema de Bayes son inusuales en los diferentes ámbitos de aplicación de la inferencia paramétrica clásica. El objetivo de este trabajo es presentar un esquema para la estimación bayesiana de parámetros bajo los supuestos de un modelo binomial. El procedimiento Bayes se estudia en comparación con la aproximación paramétrica cl??sica, ambas opciones, en su versión puntual y mediante intervalos de estimación. Se presenta también un estudio de simulación con diferentes tamaños muestrales en el que se ponen de manifiesto las ventajas del procedimiento bayesiano.AbstractThe estimation procedures based on Bayes' Theorem are still an unusual option in many of the environments of classic parametric inference. The aim of this paper is to show an effective scheme for the use of Bayesian estimation of unknown parameters. We have opted to focus on the estimation of parameters under the assumption of a binomial model, so that it can be followed by all those situations that meet the aforementioned probabilistic model. This approximation was studied in comparison with the classic parametric approximation, both in its point version and by means of interval estimation. On a study, by simulating samples of several sizes, we obtained empirical evidence regarding the advantage of the Bayesian procedure