Using ordinal optimization approach to improve efficiency of selection procedures (Q702838)

The paper deals with a problem of design selection. In particular, the aim is to choose a procedure for a best (or at least) ``acceptable'' design selection on the basis of an experience or a simulation. It is assumed that the ``optimal design'' has to be selected from a finite number of designs. Since, generally, it is known that a result of every experience or simulation obtains a noise, the results of the simulation are random ``variables''. Consequently, it is possible to try to select correct design only with a high probability. The proposed procedure takes into account not only the sample variance but also the difference of sample means when determining sample sizes. An experimental performance evaluation demonstrates the efficiency of the new procedure.