Statistics and Its Interface

Volume 12 (2019)

Number 4

A new rank sensitivity metric for decision support

Pages: 573 – 583

DOI: https://dx.doi.org/10.4310/SII.2019.v12.n4.a7

Authors

Anil Dolgun (School of Science, Mathematical Sciences, Royal Melbourne Institute of Technology, Melbourne, VIC, Australia)

Haydar Demirhan (School of Science, Mathematical Sciences, Royal Melbourne Institute of Technology, Melbourne, VIC, Australia)

Andrew Gill (Defence Science and Technology Group, Edinburgh, SA, Australia)

Dion Grieger (Defence Science and Technology Group, Edinburgh, SA, Australia)

Stella Stylianou (School of Science, Mathematical Sciences, Royal Melbourne Institute of Technology, Melbourne, VIC, Australia)

Stelios Georgiou (School of Science, Mathematical Sciences, Royal Melbourne Institute of Technology, Melbourne, VIC, Australia)

Abstract

Assessing the change in the relative performance of competing systems across a factor space generated by a combination of input variables is a common problem in decision making. We propose a new metric to assess the sensitivity of the performance rankings of a set of options when input variables are changed. The proposed metric is useful in foreseeing the impact of changing values of input variables on an output metric in complex systems through computer simulation experiments. Numerical characteristics of the proposed metric are illustrated and discussed and an application is provided to illustrate use of our metric in decision support.

Keywords

decision support systems, design of experiments, weighted Spearman footrule distance, sensitivity, performance ranking

2010 Mathematics Subject Classification

Primary 62K99, 90B50. Secondary 90C31.

Full Text (PDF format)

Received 8 June 2018

Accepted 30 April 2019

Published 18 July 2019