Shrinkage Estimation of Regression Coefficients From Censored Data With Multiple Observations
This paper considers the preliminary test and Stein-type estimation of regression parameters in exponential regression failure time distribution. We consider a situation where the lifetime data may be right censored with multiple observations taken at each regression vector. We propose improved estimators of the regression vector when it is suspected that the true regression parameter vectors may be restricted to a linear subspace. The large sample risk properties of the proposed estimators are derived. The relative merits of the proposed estimators are discussed.
KeywordsCovariance Shrinkage Expense Eter Straw
Unable to display preview. Download preview PDF.
- Ahmed, S.E. (1992). Large-sample pooling procedure for correlation. J. Roy. Statist. Soc. Ser. D 41, 425–438.Google Scholar
- James, W. and C. Stein (1961). Estimation with quadratic loss. In Proc. 4th Berkeley Symp. Math. Statist, and Prob., Vol. I, pp. 361–379. Univ. California Press, Berkeley, CA.Google Scholar
- Stein, C. (1956). Inadmissibility of the usual estimator for the mean of a multivariate normal distribution. In Proc. 3rd Berkeley Symp. Math. Statist, and Prob., Vol. I, pp. 197–206. Univ. California Press, Berkeley, CA.Google Scholar