A prediction-oriented criterion for choosing the biasing parameter in Liu estimation


Oezkale M. R., Kaciranlar S.

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS, vol.36, pp.1889-1903, 2007 (SCI-Expanded) identifier identifier

Abstract

The estimation of biasing parameter k in ridge regression is an important problem. There are many procedures in the literature for choosing k. For predictive performance, methods for selecting k were given. One of which is choosing k so as to minimize the PRESS statistic for ordinary ridge regression (Montgomery and Friedman, 1993).

Liu (1993) introduced a new biased estimator which depends on the biasing parameter d and gave some estimates of d by analogy with the estimates of k. In this article, we propose and investigate PRESS statistic for selecting d and we also give a numerical example based on widely analyzed data set on Portland cement to illustrate the results.

The estimation of biasing parameter k in ridge regression is an important problem. There are many procedures in the literature for choosing k. For predictive performance, methods for selecting k were given. One of which is choosing k so as to minimize the PRESS statistic for ordinary ridge regression (Montgomery and Friedman, 1993).