Bootstrap estimates in R

Abstract

Bootstrap estimates are based on principle of resampling with replacement. It is a compute-intensive method that presents an alternative to traditional means of unknown parameter estimation and especially their confidence intervals. Bootstrap technique allows us to estimate sampling distribution of almost any statistic based on random sampling. Besides random parameter estimation, advance can be used also in regression model statistics estimation such as coefficient of determination to compute its standard error and confidence intervals. Usage of bootstrap estimates is appropriate especially in cases when analytical solution of statistics of interest is very difficult or is not possible at all. Given high compute demands, it is essential to use appropriate statistical software for the calculation. The aim of this article is to acquaint reader with theoretical advances of bootstrap estimates creation and subsequently their calculation by usage of programming language R. More specifically, package boot, which advance is high flexibility and direct computation of confidence interval estimation without any necessity for further programming.