Title: Ensemble estimation and variable selection with semiparametric transformation models
Authors: Sunyoung Shin - University of Texas at Dallas (United States) [presenting]
Abstract: Semiparametric transformation models associate potentially time-dependent covariates on survival time. We consider a certain class of semiparametric transformation models, whose likelihood factors into separate components. When an efficient estimator of the regression parameter is available for each component, an optimal weighted combination of the component estimators, named an ensemble estimator, may be employed as an overall estimate of the regression parameter. This approach is useful when the full likelihood function may be difficult to maximize but the components are easy to maximize. Variable selection is important in such regression modelling but the applicability of existing techniques is unclear in the ensemble approach. We propose ensemble variable selection using the least squares approximation technique on the unpenalized ensemble estimator, followed by ensemble re-estimation under the selected model. We conduct numerical studies with proportional odds models to show that the proposed method outperforms alternative approaches.