[R] Some questions concerning survival tree analysis using the rpart module

Terry Therneau therneau at mayo.edu
Tue Jan 11 14:55:14 CET 2011


 All the documentation that I have on survival splitting is found in the
technical report you mention.  However, there is both a short form and a
long form of this  on our web site, did you get the larger one (52
pages)?  I admit it is not a lot.
  There are no other split algorithms implimented for survival data.  It
would be possible to add your own. Attached is a slightly updated
version of the technical report that has a section on user written split
rules.  

The pruning method is based on cost-compexity.  It is one area where
substantial improvment may be possible.  I thought a lot about this at
one time, but it's now been a decade since I put any intellectual energy
into rpart.

Terry Therneau



------begin included message ----
I succeeded in performing survival tree analysis using the “method=exp”
function, and by defining a “Surv object”. However, I wonder what kind
of splitting algorithm and pruning method are used by R to obtain the
survival tree. 

I know that for classification trees, it is possible to specify the
split algorithm using the function “parms=list(split=…)”, which can be
specified as ‘gini’ or ‘info’. Are there also multiple options for the
splitting algorithm using the “Exp” method for survival trees, that can
be defined or is there only one default option?

Is the pruning method for the survival tree in the rpart package also
based on tree cost complexity?

 

I have read the report “An introduction to recursive partitioning using
the rpart routines” that you have written. It was really helpful in
familiarizing myself with CART analysis using the rpart package.
However, I missed some detail on the background of survival tree
analysis (using rpart and in general). Do you know whether there is
another document or book that elaborates more on this subject?

--end inclusion ----
-------------- next part --------------
A non-text attachment was scrubbed...
Name: zed.pdf
Type: application/pdf
Size: 356858 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20110111/05710d10/attachment.pdf>


More information about the R-help mailing list