Title: An optimised method of weighting combination in multi-index comprehensive evaluation

Authors: Jing Ren, Yijie Xiong

Addresses: Business Administration School, Xi'an University of Technology, 710048 Xi'an, China. ' Business Administration School, Xi'an University of Technology, 710048 Xi'an, China

Abstract: The multi-index comprehensive evaluation is a key method commonly used in the research activities in industrial, agricultural, commercial and other sectors, where determining of the index weights is an essential work to be done. In this paper, an optimised method that combines the subjective and the objective weighting is proposed on the index weights aiming to find a solution to its current shortages that may be found in the multi-index comprehensive evaluation. Index in this new method for evaluation of the object is divided into two categories: with or without a large amount of data information. Furthermore, the information entropy theory and the extension AHP are used to determine the weights, respectively. Meanwhile, the extension analysis is used in synthesising the two categories of indexes, and index weights, determined by which, can be used for measurement at the same scale are also illustrated by examples of the application. This proposed method can greatly reduce the errors on the subjective evaluation, and can be easily used to determine the more objective index weight for multi-index comprehensive evaluation in different areas.

Keywords: multi-index comprehensive evaluation; extension analysis; extension AHP; analytical hierarchy process; information entropy; weighting combinations; index weights; R&D; research and development; industry; agriculture; commerce; shortages; subjective evaluation; objective weightings; applied decision sciences.

DOI: 10.1504/IJADS.2010.032239

International Journal of Applied Decision Sciences, 2010 Vol.3 No.1, pp.34 - 52

Published online: 17 Mar 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article