Applying Bootstrap Methods to Simulation Output Analysis,
Abstract
Confidence intervals obtained by bootstrap methods and normal approximation are compared, based on output data from terminating and steady-state simulations. Bootstrap intervals are equal or better than normal approximation intervals in actual probability coverages. Furthermore, bootstrap methods capture the skewness in the distribution of outputs and, therefore, are more desirable than normal approximation
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1992
- Accession Number
- ADP007112
Entities
People
- Charles B. Rea
- Chong-wei Xu
- Wei-kei Shiue
Organizations
- Southern Illinois University Edwardsville