A New Gradient Method for Solving Linear Regression Model
Norhaslinda Zull1, Nurul ‘Aini2, Mohd Rivaie3, Mustafa Mamat4
1Norhaslinda Zull, Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin, Tembila, Terengganu, Malaysia.
2Nurul ‘Aini, Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, Segamat, Johor, Malaysia.
3Mohd Rivaie, Department of Computer Science and Mathematics, Universiti Teknologi MARA, Kuala Terengganu, Terengganu, Malaysia.
4Mustafa Mamat, Faculty of Informatics and Computing, Universiti Sultan Zainal Abidin, Tembila, Terengganu, Malaysia.
Manuscript received on 16 February 2019 | Revised Manuscript received on 07 March 2019 | Manuscript Published on 08 June 2019 | PP: 624-630 | Volume-7 Issue-5S4, February 2019 | Retrieval Number: E11300275S419/19©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: One of the commonly used optimization methods is the conjugate gradient (CG) method. This method is highly practical for solving large scale problems and applicable for real life. This study suggests another CG method that fulfills the sufficient descent and global convergence properties. The robustness and efficiency of the proposed method are evaluated by comparison with other established CG methods. The numerical testing uses sixteen test functions in MATLAB subroutine programming under strong Wolfe line search. Numerically, the result concludes that the new CG method has the best performance in term of iteration number (NOI) and CPU time. This method is then implemented for solving linear regression model in order to show its applicability. Hence, this method has been proven to be successful.
Keywords: Conjugate Gradient Method, Global Convergence, Regression Analysis, Strong Wolfe.
Scope of the Article: Data Mining Methods, Techniques, and Tools