July 24th 2012
Faculty of Economic and Business Sciences | Room 145
Model building in non proportional hazard regression
2012/07/24 - 11:45 h | Mar Rodríguez Girondo, University of Vigo
Variable selection and model choice are important issues in non proportional hazard regression. Recent developments of statistical methods allow for a flexible modeling of the variables affecting survival, including the inspection of possible time-dependent associations. Despite their immediate appeal in terms of flexibility, these models introduce additional difficulties when a subset of covariates and the corresponding modeling alternatives have to be chosen. However, a piecewise exponential representation of the original survival data enables to use a Poisson likelihood based estimation scheme. We propose to conduct such data transformation and adapt model building procedures proposed in generalized additive models regression settings to the survival context. Three different model building technics are adapted and compared via an intensive simulation study. An application to prognosis after discharge for patients who suffered an acute myocardial infarction is presented. This is joint work with Thomas Kneib, Carmen Cadarso-Suárez, and Emad Abu-Assi.
Goodness-of-fit tests for a semiparametric model under random double truncation
2012/07/24 - 13:30 h | Carla Moreira, University of Vigo
Doubly truncated data may appear in Survival Analysis, astronomy, or economy, among other fields. The semiparametric estimator of a doubly truncated random variable is based on a parametric specification of the truncation times distribution function. This semiparametric estimator outperforms the NPMLE when the parametric information is correct. In this paper we introduce several goodness-of-fit tests for the semiparametric model. The several tests are investigated through simulations. For illustration purposes, a real data application is given. This is joint work with Jacobo de Uña-Álvarez and Ingrid Van Keilegom.
A nonparametric ANOVA-type test for regression curves based on characteristic functions
2012/07/24 - 13:15 h | Juan Carlos Pardo-Fernández, University of Vigo
This article studies a new procedure to test for the equality of k regression curves in a fully nonparametric context. The test is based on the comparison of empirical estimators of the characteristic functions of the regression residuals in each population. The asymptotic behaviour of the test statistic is studied in detail. It is shown that under the null hypothesis the distribution of the test statistic converges to a combination of chi-square random variables. Under certain restrictions on the populations, the asymptotic null distribution of the test statistic is a chi-square with k-1 degrees of freedom. The practical performance of the test based on the asymptotic null distribution is investigated by means of simulations. This is joint work with María Dolores Jiménez Gamero (Universidad de Sevilla) and Anouar El Ghouch (Université catholique de Louvain).