A new maximum relevance-minimum multicollinearity (MRmMC) method for feature selection and ranking. (July 2017)

• Record Type: Journal Article
• Authors: Senawi, Azlyna; Wei, Hua-Liang; Billings, Stephen A.
• Is Part Of: Pattern recognition. Volume 67(2017:Jul.)
• Page Start: 47
• Page End: 61
• Publication Date: 2017-07
• Subjects: Dimensionality reduction -- Feature selection -- Classification -- Correlation measure -- Qualitative and quantitative variables; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2017.01.026
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A new maximum relevance–minimum multicollinearity (MRmMC) method is proposed. The proposed MRmRC algorithm was applied to a number of real-life datasets; experimental results are reported and compared with several state-of-the-art methods. Numerical analysis results confirmed the promising performance of the proposed method. Abstract: A substantial amount of datasets stored for various applications are often high dimensional with redundant and irrelevant features. Processing and analysing data under such circumstances is time consuming and makes it difficult to obtain efficient predictive models. There is a strong need to carry out analyses for high dimensional data in some lower dimensions, and one approach to achieve this is through feature selection. This paper presents a new relevancy-redundancy approach, called the maximum relevance–minimum multicollinearity (MRmMC) method, for feature selection and ranking, which can overcome some shortcomings of existing criteria. In the proposed method, relevant features are measured by correlation characteristics based on conditional variance while redundancy elimination is achieved according to multiple correlation assessment using an orthogonal projection scheme. A series of experiments were conducted on eight datasets from the UCI Machine Learning Repository and results show that the proposed method performed reasonably well for feature subset selection.