An Examination of the Performance of Two-Stage Group Screening for use in Computer Simulation.
Abstract
When only a limited number of simulation runs are available, it is usually impossible to thoroughly investigate all of the factors under consideration. Often, though, it is anticipated that only a small subset of the original factors is important. Accordingly, it is desired to screen the factors in order to help identify those that do exert an appreciable effect on the simulation response. Two-stage group screening is one possible strategy that may be used. However, a basic assumption of this strategy is that the directions of possible effects are known, a priori. This report examines, in the case of zero error variance (i.e., when the simulation response is observed without random error), the performance of two-stage group screening when the assumption of known directions is relaxed. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1980
- Accession Number
- ADA091553
Entities
People
- Carl A. Mauro
- Dennis E. Smith