Survival tree

The parametric hazards regression model is sometimes seen as a restrictive class of models by practitioners, and the Cox proportional hazards regression is sometimes preferred over its parametric counterpart. Compared with the parametric models, the interpretation is sometimes lost, and the regular practitioner finds it difficult to connect with the hazards regression model. Of course, an alternative is to build a survival tree over the pseudo observations. Such an attempt can be seen in Tattar's (2016) unpublished paper. Gordon and Olshen (1985) made the first attempt to build a survival tree and many scientists have continued constructing it. LeBlanc and Crowley (1992) are among the most important contributors to set up a survival tree. Zhang and Singer (2010) have also given a systematic development of related methods, and chapters 7-10 of their book deal with survival trees. The basic premise remains the same, and we need good splitting criteria in order to create...