A new three-term conjugate gradient method with application to regression analysis

Nur Idalisa (Universiti Teknologi MARA)
Mohd Rivaie (FUniversiti Teknologi MARA)
Nur Hidayah Mohd Noh (Universiti Teknologi MARA)
Mohd Agos Salim Nasir (Universiti Teknologi MARA)
Nurul Hafawati Fadhilah (Universiti Teknologi MARA)
Norma Alias (Universiti Teknologi Malaysia)

International Journal of Electrical and Computer Engineering
Vol 12, No 5: October 2022

Conjugate gradient (CG) method is well-known for its ability to solve unconstrained optimization (UO.) problems. This article presenting a new CG method with sufficient descent conditions which improves the former method developed by Rvaie, Mustafa, Ismail and Leong (RMIL). The efficacy of the proposed method has been demonstrated through simulations on the Kijang Emas pricing regression problem. The daily data between January 2021 to May 2021 were obtained from Malaysian Ministry of Health and Bank Negara Malaysia. The dependent variable for this study was the Kijang Emas price, and the independent variables were the coronavirus disease (COVID-19) measures (i.e., new cases, R-naught, death cases, new recovered). Data collected were analyzed on its correlation and coefficient determinant, and the influences of COVID-19 on Kijang Emas price was examined through multiple linear regression model. Findings revealed that the suggested technique outperformed the existing CG algorithms in terms of computing efficiency.