[R] Lapack routine dgesv: system is exactly singular

[Terry Therneau]

I was sent a copy of the data on request.  A quick look shows that 

> range(days.alive[censored==0])
[1]    0 1825
>  range(days.alive[censored==1])
[1] 1826 1826

 The original call of survdiff(Surv(days.alive, censored) ~ group) will
assume that censored=1 corresponds to deaths and 0 to alive; from the
above this is almost certainly backwards.  If instead we force 0 to be
the "yes they're dead" group the results look better.

> survdiff(Surv(days.alive, censored==0) ~ group, f2)
Call:
survdiff(formula = Surv(days.alive, censored == 0) ~ group, data = f2)

                          N Observed Expected (O-E)^2/E (O-E)^2/V
group=PRI_CAS_5_NODU   3326     1129     1745     217.7       281
group=SEC_CAS_5_NODUP 13469     6731     6115      62.1       281

 Chisq= 281  on 1 degrees of freedom, p= 0 

In the original setup the test statistic had value 0 + roundoff error
and std = 0 + roundoff error due to all the "deaths" being tied on
exactly the same day, and you were getting the matrix version of a 0/0
error.

  Terry T

------- begin included message ----
I have a problem with this error, I have searched the archives and found
previous discussion about this, can I cannot understand how the
explanations
apply to what I am trying to do.

 

I am trying to do Log_rank Survival analysis, I have included tables and
str command, is it a factor/integer problem? If so how do I correct
this, as all my attempt to recode the data have failed.

 

> survdiff(Surv(f2$days.alive , f2$censored)~group, data=f2)

Error in drop(.Call("La_dgesv", a, as.matrix(b), tol, PACKAGE =
"base")) : 

  Lapack routine dgesv: system is exactly singular