In Section 2.2.2 the integration of the Cox and Weibull model with adjustment for measurement error into the MOB algorithm is derived theoretically. For the implementation the focus is set to the Weibull model. In the R package party (Hothorn et al., 2010) a function mob() (Zeileis et al., 2008) is available to conduct model-based recursive partitioning. A detailed vignette for mob() is available in Zeileis et al. (2010). A model-based parameter estimation is already integrated for the Weibull model. The main challenge of the thesis is to extend the implemented estimation for the Weibull model with measurement error correction. For the implementation R version 2.11.1 (R Development Core Team, 2010) is used. The functionality of mob() and the extended implementation are described in the following.
KeywordsChild Node Current Node Moment Generate Function Weibull Model Parameter Instability
Unable to display preview. Download preview PDF.