Modification of search direction in steepest descent method and its application in regression analysis

Abstract

Steepest descent (SD) method is the basic and simple algorithm for minimizing function of n variables. Although this method is said to lead to very slow convergence, the small change of the search direction for each line can improved the method. Therefore, in this paper, we propose new steepest descent (SD) method which focused on the modification of the direction that possesses sufficient descent conditions and global convergence properties as our first objective. The second approach is that we present the implementation of the proposed modification of SD method into the regression analysis for the real life problems. © 2020, World Academy of Research in Science and Engineering. All rights reserved.