A new multidimensional model II regression based on bisector approach

Abstract

A new multidimensional Model II regression based on bisector point of view (BRM-II) is introduced for multivariate problems that may contain measurement error. The suggested method is constructed depending on using the bisector of the minor angle between two hyperplanes identified by linear regression. The performance of the proposed method are examined by simulations up to ten variables for different sample sizes and distribution types in terms of the Mean Square Error. Moreover, the BRM-II is applied to two real problems with two and three variables, and compared with the existing methods. The results indicate that the BRM-II is easy applicable and offers relatively better accuracy. The relevant method can be easily coded in any programming language provides convenience in its application. Thus, the proposed method provide powerful tool for prediction of relevant real life problems.