2
$\begingroup$

I have left censored data where the distribution is known (it's near enough lognormal, at least in theory).

I'd like to calculate some simple summary stats: geometric mean and standard deviation in this case.

I've previously used R's NADA package for this but it is no longer on CRAN. Is there an alternative available?

EDIT:

I contacted Lopaka Lee, the package maintainer, and he says that he's

working on updating the package for the upcoming release of R

so hopefully the package's absence is only temporary.

Improve this question
$\endgroup$
0

1 Answer 1

Reset to default
4
$\begingroup$

I will let other suggest better alternatives to NADA, but it seems the package is still available on CRAN, in the Archive section. The last version is from May, 2009.

Installation went fine for me, using 

R CMD install NADA_1.5-2.tar.gz

Under Windows, I guess you can just download the tgz and use built-in install facilities.

Improve this answer
$\endgroup$
3
  • 1
    $\begingroup$ (+1) I located and installed this package less than two months ago. For more sophisticated work, consider using censored regression models. $\endgroup$
    – whuber ♦
    Commented Dec 7, 2010 at 17:10
  • $\begingroup$ @whuber: I've done MCMC modelling on the data but the target audience is chemists so some simpler stats are useful to warm them up. My worry is that NADA was withdrawn because it gave the wrong answers. (I know it's calculations of medians was dodgy.) $\endgroup$
    – Richie Cotton
    Commented Dec 7, 2010 at 17:35
  • $\begingroup$ @Richie That is extremely interesting to hear. I didn't test NADA that far: I tried it out, quickly found it lacking in capabilities, and went immediately to survival analysis and censored regression models. $\endgroup$
    – whuber ♦
    Commented Dec 7, 2010 at 18:48

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.