# [R] Kaplan-Meier survfit problem

David Winsemius dwinsemius at comcast.net
Mon Dec 12 02:01:20 CET 2011

```On Dec 11, 2011, at 5:48 PM, Esteban Cervetto wrote:

> 2011/12/11 Esteban Cervetto <estebancster at gmail.com>
>
>> I am working with uncensored data.
>>
>> I have duration of workers compensation. Then I have for each the
>> number
>> of days thet it doesn't work.
>> This sample is not censored at right because I query only work
>> accidents
>> with date of return of work (saned)
>>
>> That is because I have only one vector: the number of days that the
>> worker
>> doesn't work.
>> Reading works that uses this library, noticed that it need's a
>> vector to
>> mark the type of termination. That is because I did a formula  Y(x)
>> = 1
>> as.numeric(T.**201110))~1
>>
>> Your words were inspired me to do this: I solvet it putting to T.
>> 201110
>> this proper vector and a vector of ones.
>>
>
>    T.201110\$censor <- apply(T.201110,1,function(row) 1)    ##is
> there a
> best method to do that? it takes much time
>
>    I believe that this vector is superfluous, because the result of
> the
> formula is ever 1

You may be correct. The help page for Surv says "Although unusual, the
event indicator can be omitted, in which case all subjects are assumed
to have an event."

>
>
>> km1<-survfit(Surv(T.201110\$dias,T.201110\$censor)~1)
>>
>>
>
>
>> 2011/12/10 David Winsemius <dwinsemius at comcast.net>
>>
>>>
>>> On Dec 10, 2011, at 6:39 PM, capitantyler wrote:
>>>
>>> done it, again, i have the next problem
>>>>
>>>> "The object (list) cannot be corced as "double"
>>>>
>>>> Original:
>>>> *km1 <- survfit(Surv(as.numeric(T.**201110))~1)*

If you are adding the "**" for emphasis, it is certainly confusing my
understanding of what your original code was, which i'm now wondering
you ever provided. Perhaps it was the lack of a data argument. Hard to
tell.

This "works":

> fit <- survfit(Surv(time) ~ 1, data = aml[aml\$status==1, ])
> plot(fit)

--
David

>>>> Error en Surv(as.numeric(T.201110)) :
>>>> el objeto (list) no puede ser coercionado a 'double'
>>>>
>>>
>>> I do not read that language, but I am surprised to see a single
>>> vector
>>> being used as an argument to Surv(). When I use Surv(. , .) it is
>>> with two
>>> vectors, an interval and a censor variable.
>>>
>>>
>>>
>>>> note that need it convert to numeric class, otherwise:
>>>>
>>>> *km1 <- survfit(Surv((T.201110))~1)*
>>>> Error en Surv((T.201110)) : Time variable is not numeric
>>>> --
>>>> View this message in context: http://r.789695.n4.nabble.com/**
>>>> Kaplan-Meier-survfit-problem-**tp2015369p4181476.html<http://r.789695.n4.nabble.com/Kaplan-Meier-survfit-problem-tp2015369p4181476.html
>>>> >
>>>>
>>>
>>> Context? Yes. We do want context, but we don't want to go to no
>>> steenking
>>> Nabble.
>>>
>>>
>>> Sent from the R help mailing list archive at Nabble.com.
>>>>
>>>> Arrrgh. The rhelp mailing list is NOT at Nabble. Nabbel si a
>>>> commercial
>>> mirror of the real thing. And it isn't really an archive. either,
>>> since
>>> they start discarding posts after a year or two.
>>>
>>> --
>>> David Winsemius, MD
>>> West Hartford, CT
>>>
>>>
>>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help