The answers to your questions consume an entire chapter of my book, so it is difficult to write a reasonable email response. When there are time dependent covariates, the creation and interpretation of a survival curve is particularly subtle. The survfit.coxph routine calculates the usual estimates: Breslow, Fleming-Harrington, or Kalbfleisch-Prentice; these differ in how they handle ties but the numerical results are usually very close. A treatise on the formulas, theory, and derivation of these is beyond the scope of this message.
Terry Therneau --- begin included message --- I am in an urgent using R to fit the cox proportional model. My data is a data frame including 100 individuals which are software stress tests. There are time-to-failure and six covariates meaning the system resource (every 4 minutes). Is it possible to use R to fit the Cox model? I used "*coxph*" to estimated the coefficients for both time-independent and time-dependent covariates. And next I need to calculated the survival and hazard rate for cox model with time-independent and that with time-dependent covariates. I see many people use *survfit.coxph *to get the survival. Could anybody please tell me how is the survival value calculated? For example, what kind of procedure. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.