0
$\begingroup$

I would like to manually compute the lognormal distribution but I am not clear on how to compute the CDF value that is required in the formula for the Survival function.

The formula is as below:

$S(x)=1−Φ(\frac{ln(x)}{σ})$

Is there a way to get this value using R?

Thanks

$\endgroup$
5
  • $\begingroup$ so far what I have done is to use this formula in R: S(t) = 1 - plnorm(log(t) / model$scale) . Is this correct? $\endgroup$
    – MugB
    Nov 13, 2019 at 13:16
  • $\begingroup$ Why doesn't $\mu$ appear in your formula? $\endgroup$
    – Glen_b
    Nov 14, 2019 at 2:56
  • $\begingroup$ okay i found this link: stats.stackexchange.com/questions/33664/… with another formula. Does this mean that the μ comes from the survreg function? $\endgroup$
    – MugB
    Nov 14, 2019 at 9:21
  • $\begingroup$ it shows from this link: blogs2.datall-analyse.nl/2016/02/17/… that the mu is taken from coef of the survreg function. Is this correct? $\endgroup$
    – MugB
    Nov 14, 2019 at 12:57
  • 1
    $\begingroup$ Sorry, wasn't able to post this yesterday -- I eventually figured out that the point of the non-mu formula is just as a "standard" form of lognormal with median 1, which you might treat as a multiplicative error term. In that case, yes, the mean survival (conditional on the predictors) would supply the mu that this would represent the variation around. $\endgroup$
    – Glen_b
    Nov 15, 2019 at 13:44

1 Answer 1

1
$\begingroup$

I have found the solution for the formula when coding it in R and it is as below

$S(t) = 1 - plnorm(t, meanlog, sdlog)$

both the meanlog and sdlog value can be derived from computing the shape and the scale from the survreg results.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.