Bootstrap não paramétrico aplicado ao modelo de regressão linear múltipla com uso da planilha do excel

Abstract

The purpose of regression analysis is to study the relationship between a variable Y, called a dependent or response, and one or more of independent or regressing variables. That is, to understand how certain variables influence the behavior of another variable. It is understood that the main objective of regression is to provide an equation that relates the dependent variable with the independent variables considered in the model, also making it possible to make predictions about the behavior of the studied phenomenon, adjust parameters or the model and make inferences about them. Many applications of regression analysis involve situations in which there is more than one regressor variable. In this work we apply the Bootstrap method in pairs in the use of the tests of significance required in the multiple regression model. The objective was to relax the need for the error component (∈) of the model to have normal distribution and constant variance. The procedure was implemented in the Excel spreadsheet and did not require programming. All the necessary functions were described in detail for use in a business environment using the Excel spreadsheet, also favoring the understanding of the bootstrap method.