|
Generation of Correlated Random Variables
One of the emerging areas of
research of interest to numerical studies that involve simulation is the ability
to generate random variables with various statistical distributions and with a
desired correlation structure. The most commonly used method to achieve this is
based on the algorithm proposed by Iman and Conover (1982), which requires
desired correlation structure to be specified as the Spearman rank order
correlation matrix. A new method has been proposed and tested. The new method
has a number of fine-tuning parameters that can help the modeler achieve the
desired accuracy. Publication detailing this work will eventually appear in the
European Journal of Operational Research.
Reference: Iman, R. and W.
Conover. 1982. A Distribution Free Approach to Inducing a Rank Correlation among
Input Variables. Communications in Statistics – Simulation and Computation, 11
(3) p. 311-334.
|