Estimador regressão para dados assimétricos

Abstract

This paper aims at proposing a modification in the structure of the regression estimator, considering the estimator of the coefficient of linear regression line, ie, denoted by hat beta skewness rather than under normality. Thus we named this proposed estimator as an estimator for skewness data. From this change we will compare the estimators on the bias and variance of the same, the variance of hat beta estimates for each estimator, assuming a normal distribution skewed. To illustrate the proposed methodology, we consider a simulation study and application to actual data.