Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/87358
Title: On weighted least squares estimation for parameters of the two-parameter weibull distribution
Authors: Zhang, L.F.
Xie, M. 
Tang, L.C.
Keywords: Mean square error
Parameter estimation
Weibull distribution
Weighted least squares estimation
Issue Date: 2006
Citation: Zhang, L.F.,Xie, M.,Tang, L.C. (2006). On weighted least squares estimation for parameters of the two-parameter weibull distribution. 2006 Proceedings - 12th ISSAT International Conference on Reliability and Quality in Design : 318-322. ScholarBank@NUS Repository.
Abstract: This paper presents an alternative method for calculating weights to be used in weighted least squares estimation (WLSE) technique for estimating the two Weibull parameters. As a common practice, weights are calculated by the reciprocals of the variances of predictor variable values. The existing WLSE methods including Bergman [10], Faucher and Tyson [11], Hung [12] and Lu et al. [13] use approximated values of the variances to calculate weights. In fact, the exact values of the variances of predictor variable values can be deducted through analytical analysis. The present paper describes the method for deducing the exact values of the variances, and also provides an approximation formula to simplify the calculation. Step-by-step procedures are provided for the proposed WLSE technique. Simulation results show that for estimating the shape parameter, the proposed procedure is more accurate than the existing WLSE methods and always generates smallest mean square error (MSE).
Source Title: 2006 Proceedings - 12th ISSAT International Conference on Reliability and Quality in Design
URI: http://scholarbank.nus.edu.sg/handle/10635/87358
ISBN: 0976348616
Appears in Collections:Staff Publications

Show full item record
Files in This Item:
There are no files associated with this item.

Page view(s)

36
checked on Nov 9, 2018

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.