A mathematical programming approach to sample coefficient of variation with interval-valued observations
-
Shiang-Tai Liu [1]
-
[1] Vanung University
Vanung University
Taoyuan District, Taiwán
-
- Localización: Top, ISSN-e 1863-8279, ISSN 1134-5764, Vol. 24, Nº. 1, 2016, págs. 1-18
- Idioma: inglés
- Enlaces
-
Resumen
The coefficient of variation is a useful statistical measure, which has long been widely used in many areas. In real-world applications, there are situations where the observations are inexact and imprecise in nature and they have to be estimated. This paper investigates the sample coefficient of variation (CV) with uncertain observations, which are represented by interval values. Since the observations are interval-valued, the derived CV should be interval-valued as well. A pair of mathematical programs is formulated to calculate the lower bound and upper bound of the CV. Originally, the pair of mathematical programs is nonlinear fractional programming problems, which do not guarantee to have global optimum solutions. By model reduction and variable substitutions, the mathematical programs are transformed into a pair of quadratic programs. Solving the pair of quadratic programs produces the global optimum solutions and constructs the interval of the CV. The given example shows that the proposed model is indeed able to help the manufacturer select the most suitable manufacturing process with interval-valued observations.
