CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Boosting methods for effects selection in Cox frailty models Authors:  Andreas Groll - Technical University Dortmund (Germany) [presenting]
Trevor Hastie - Stanford University (United States)
Thomas Kneib - University of Goettingen (Germany)
Gerhard Tutz - Ludwig-Maximilians-University Munich (Germany)
Abstract: As in many other sorts of regression problems, also in survival analysis it has become more and more relevant to face high-dimensional data with lots of potentially influential covariates. These generally can have time-constant or time-varying effect types, which a priori is often unknown to the modeler. A possible solution is to apply estimation methods that aim at the detection of the relevant effect structure by using regularization methods. The effect structure in the Cox frailty model, which is the most widely used model that accounts for heterogeneity in survival data, is investigated. Since in survival models one has to account for possible variation of the effect strength over time, the selection of the relevant features has to distinguish between several cases: covariates can have time-varying effects, can have time-constant effects or be irrelevant. To address these model selection issues, a likelihood-based component-wise boosting approach is proposed that is able to distinguish between these types of effects and one obtains a sparse representation that includes the relevant effects in a proper form. The method is applied to a real world data set, illustrating that the complexity of the influence structure can be strongly reduced by using the proposed boosting approach.