A POISSON PROCESS APPROXIMATION FOR GENERALIZED K–S CONFIDENCE REG |
| |
Authors: | H Arsham DR Miller |
| |
Institution: | Dept. of Information and Quantitative Sciences School of Business University of Baltimore Baltimore, Maryland 21201 U.S.A;Dept. of Operations Research School of Engeneering and Applied Science The George Washington University Washington, D.C. 20052 U.S.A. |
| |
Abstract: | One–sided confidence regions for continuous cumulative distribution functions are constructed using empirical cumulative distribution functions and the generalized Kolmogorov–Smimov distance. The band width of such regions becomes narrower in the right or left tail of the distribution. To avoid tedious computation of confidence levels and critical values, an approximation based on the Poisson process is introduced. This approximation provides a conservative confidence region; moreover, the approximation error decreases monotonically to 0 as sample size increases. Critical values necessary for implementation are given. Applications are made to the areas of risk analysis, investment modelling, and analysis of fault–tolerant systems. |
| |
Keywords: | One—sided confidence regions for continuous cdf Empirical cdf Generalized K—S statistics Poisson process Risk analysis (assessment) Investment modeling Fault—tolerant systems |
|
|