Application of bootstrap techniques in regression models

Due to the nature of multiple linear regression, the least squares estimation of regression parameters when explanatory variables are linearly related tends to be in the wrong sign and has large variance. To eliminate these kinds of abnormal effects, there are some useful techniques which reduce or totally eliminates this problem. Biased estimation methods are examples of these methods. In fact, this study aims to identify and observe the usefulness of these methods, particularly the bootstrap method. Furthermore, this study also aims to identify the best the k selection method while determining confidence interval levels. The advantage of using the bootstrap method is that this method enables the researcher to produce new data set from the existing one via resampling it. On the other hand, k selection methods, which aim to minimize the mean square errors of the ridge estimators, have drawbacks when compared to the bootstrap method. The purpose of this study is to identify the existing literature on multiple linear regression via both ridge estimator and bootstrap, provide a comprehensive bootstrap study and simulations via considering the k selection method and its effects on confidence interval levels.