Dienstag, 19. Oktober 2010, 11-12 Uhr - Raum: W9-109

Javier Trejos
Escuela de Matemática, Universidad de Costa Rica

Genetic Algorithm for Variable Selection in Linear Regression

We study the application of a genetic algorithm in the problem of variable selection for multiple linear regression, minimizing the least squares criterion. The algorithm is based on a chromosomic representation of variables that are considered in the least squares model. A binary chromosome indicates the presence (1) or absence (0) of a variable in the model. The fitness function is based on the adjusted square R, proportional to the fitness for chromosome selection in a roulette wheel model selection. Usual genetic operators, such as crossover and mutation are implemented. Comparisons are performed with benchmark data sets, obtaining satisfying and promising results.

 

Dienstag, 16. November 2010, 11-12 Uhr - Raum: W9-109

Prof. Dr. Ostap Okhrin
Humboldt-Universität zu Berlin

Summary on Hierarchical Archimedean Copulas

This talk consists of several papers on the hierarchical Archimedean copulas. At first we discuss the methods of determining the structure and estimation of this copula family. Afterwards we analyse the properties of hierarchical Archimedean copulas. This class is a generalisation of the Archimedean copulas and allows for general non-exchangeable dependency structures. We show that the structure of the copula can be uniquely recovered from all bivariate margins. We derive the distribution of the copula value, which is particularly useful for tests and constructing confidence intervals. Furthermore, we analyse dependence orderings, multivariate dependence measures and extreme value copulas. Special attention we pay to the tail dependencies and derive several tail dependence indices for general hierarchical Archimedean copulas.

 

Dienstag, 14. Dezember 2010, 11-12 Uhr - Raum: W9-109

Prof. Philippe Lambert
Université de Liège

An additive location-scale model for interval-censored data

An additive model for the location and dispersion of a continuous response with an arbitrary smooth conditional distribution is proposed. B-splines are used to specify the three components of the model. It can be extended to deal with interval censored data and multiple covariates. As an illustration, the relation between age, the number of years of full-time education and the net income (provided as intervals) available per person in Belgian households is studied from survey data.

 

Dienstag, 18. Januar 2011, 11-12 Uhr - Raum: W9-109

Prof. Dr. Philipp Sibbertsen
Leibniz Universität Hannover

An identification problem in ESTAR models and a new model

In ESTAR models it is usually difficult to determine parameter estimates. We show that the phenomena of getting strongly biased estimators is a consequence of an identification problem, the problem of properly distinguishing the transition function in relation to extreme parameter combinations. This happens in particular for either very small or very large values of the error term variance. Furthermore, we introduce a new alternative model -the T-STAR model- which has similar properties as the ESTAR model but reduces the effects of the identification problem. We also derive a linearity and a unit root test for this model.