1
Simulation study on the performance of popular methods of parameter estimation based on jackknife and bootstrap resamplings | |
Author | Chieng, Ray Fong |
Call Number | AIT Thesis no.CS-89-10 |
Subject(s) | Lognormal distribution |
Note | A thesis submitted in partial fulfilment of the requirements for the degree of Master of Engineering |
Publisher | Asian Institute of Technology |
Abstract | Two new methods, Jackknife and Bootstrap, have been proposed to acquire the bias and variance of any estimator. Traditionally, the variance of an estimator is obtained by the asymptotic theory, which is sometimes difficult to apply because it requires the analytical derivation of the needed formulas. In this study, extensive simulation studies were made to investigate the performance of these different approximation methods. For this purpose, the T-year event estimator was used, along with three popular distributions, namely the three- parameter lognormal, the generalized extreme value and the three-parameter gamma distributions. The number of replications in simulation runs was set equal to 1000 for the asymptotic theory and 500 for the Jackknife and Bootstrap methods. The range of sample size was from 20 to 100. The "true value" of the variance of xT is assumed to be the variance of xT which is acquired from all the replicative estimates of xT by any method considered. For the method of maximum likelihood the variance estimator of xT can be derived from the Fisher information matrix by the asymptotic theory. For other methods of parameter estimation, Jackknife and Bootstrap results were obtained. By examining the simulation results it is difficult to decide which method is completely appropriate to these three distributions. Nevertheless, it was found that the Jackknife and Bootstrap methods are good in many cases, and the asymptotic theory can produce comparable results in several situations. |
Year | 1989 |
Type | Thesis |
School | School of Engineering and Technology (SET) |
Department | Department of Information and Communications Technologies (DICT) |
Academic Program/FoS | Computer Science (CS) |
Chairperson(s) | Huynh, Ngoc Phien; |
Examination Committee(s) | Nagarur, Nagendra N.;Gonzales, R.L.; |
Scholarship Donor(s) | R.O.C.; |
Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1989 |