Sains Malaysiana - 2025 : Improved Robust Principal Component Analysis based on Minimum Regularized Covariance Determinant for the Detection of High Leverage Points in High Dimensional Data

MIDI, HABSHAH and JAAZ SUHAIZA and MOHD ASLAM and HANI SYAHIDA (2025) Sains Malaysiana - 2025 : Improved Robust Principal Component Analysis based on Minimum Regularized Covariance Determinant for the Detection of High Leverage Points in High Dimensional Data. Sains Malaysiana, 54 (8). pp. 2087-2097. ISSN 0126-6039

Text Improved Robust Principal Component Analysis based on Minimum Regularized Covariance Determinant for the Detection of High Leverage Points in High Dimensional Data.pdf Download (931kB)

Abstract

This paper presents an improved Robust Principal Component Analysis (IRPCA) method to enhance the detection of high leverage points (HLPs) in high-dimensional data. The proposed approach combines Principal Component Analysis (PCA) for dimensionality reduction with the Minimum Regularized Covariance Determinant (MRCD) estimator and employs Robust Mahalanobis Distance (RMD) for identifying HLPs. Simulation studies and real data applications demonstrate that IRPCA achieves accurate detection with no masking effect, minimal swamping effect, and significantly faster computational time than ROBPCA and MRCD-PCA, making it an efficient diagnostic tool for high-dimensional data analysis.

Item Type:	Article
Uncontrolled Keywords:	High Leverage Point; Minimum Regularized Covariance Determinant; Principal Component Analysis; Robust Mahalanobis Distance; High Dimensional Data; IRPCA
Subjects:	Q Science > QA Mathematics
Divisions:	Institute of Graduate Studies (IGS)
Depositing User:	LIBRARY2 UPTM
Date Deposited:	30 Jul 2026 07:59
Last Modified:	30 Jul 2026 07:59
URI:	http://eprints.uptm.edu.my/id/eprint/5930

Actions (login required)

View Item