On Robustness of Resampling-Estimators for Linear Regression Models
Proceedings of the International Symposium on Stochastic Models in Reliability, Safety, Security and Logistics
2005
Aleksandrs Andronovs,
Jeļena Fiošina
The application of the resampling-approach to the linear regression model analysis is considered. The mean square estimators are known to be the best estimators in case of regression model without nuisance observations. Alternatively, in the case of model with nuisance observations called disturbed model the classical approach gives bad, biased estimators. The considered numerical example shows that the resampling-approach gives comparably good results for disturbed models.
Atslēgas vārdi
linear regression models, resampling methods
Andronovs, A., Fiošina, J. On Robustness of Resampling-Estimators for Linear Regression Models. No: Proceedings of the International Symposium on Stochastic Models in Reliability, Safety, Security and Logistics, Izraēla, Beer Sheva, 15.-17. februāris, 2005. Beer Sheva: Sami Shamoon College of Engineering, 2005, 6.-11.lpp. ISBN 9984668797.
Publikācijas valoda
English (en)