WEIGHTING METHODS IN THE CONSTRUCTION OF AREA DEPRIVATION INDICES

Abstract

This study applies and compares several weighted average (WA) methods and Principal Component Analysis (PCA) for the construction of composite area-based deprivation index. The WA methods are based on weights that depend on standard deviation, correlation and data entropy. This paper also proposes three new approaches of WA method by suggesting their respective weights to depend on mean absolute deviation, inter-quartile range and data entropy where the probability is estimated by empirical density function and Gaussian kernel function. The deprivation indices produced by WA methods and PCA are then utilized to rank deprivation level of eighty-one administrative districts in Peninsular Malaysia.