# How to run survival analysis on big dataset?

I am recently involved in a project that needs to analyze the survival time of objects. Therefore, I plan to use the rms package to build a Cox model. The problem is, since the dataset I have is so big (about 450,000 instances, and each instances is associated with 9 covariables), that the R environment is fail to handle this. Anyone suggestions?

-
maybe this question is more suitable for stackoverflow? –  RJ- May 25 '12 at 6:22
I am not sure this is about the limitation of R or about the time complexity of the Cox model. Therefore I put it here. –  Yoanh27 May 25 '12 at 6:32
I do not have an answer for you... However, I think you should clarify whether the problem arises when your model is fitted or if it arises sooner (e.g., at the importation step)? Also, do you have any reason for not using 'coxph' from the survival package? –  ocram May 25 '12 at 7:50
@ocram Thanks for your reply. I using the rms package because the validation function provided there can output the Dxy measure directly. No, the dataset can be imported smoothly, and I can even "summary(data)". So, I think it should be related to the complexity of the Cox algorithm. Is there any conclusion or report mentioned the scale ability of Cox model? Thank you –  Yoanh27 May 25 '12 at 7:59
Does working with all 450,000 cases really figure to give you much more useful information than analyzing a random subset of, say, 20,000? –  rolando2 May 27 '12 at 14:00
You can try the survival package in R-core.
library(survival)

Not sure how it compares to rms.