Nicholas,
You're making me depressed!
Cody Hamilton, PhD
Edwards Lifesciences
Hi,
I just saw this thread. This issue, and the larger scale issue of open
source in industry
is being addressed. One has to realize that the behemoth that is the
clinical aperatus
of the pharma industry is very
Greg,
This is very interesting. Perhaps something similar could be worked out
here. Do you have to get MS Word users to work only with the template you
provide, or can they provide you any old MS Word document?
Regards, -Cody
Cody Hamilton, PhD
Edwards Lifesciences
But sweave is expanding.
I should have also noted that Sweave is available for use with R. This is
offset, however, by the fact that I will probably never be able to convince
anyone to use Latex. This is a pity as I often find myself admiring
reports done in Latex as opposed to the ones I have worked on in MS Word.
Alain,
Can you tell us what you plan to use R for?
Regards,
-Cody
[EMAIL PROTECTED] a écrit :
I am new to using R and would appreciate some advice on
which books to start with to get up to speed on using R.
My Background:
1-C# programmer.
2-Programmed directly using IMSL (Now Visual
Following up to some extent on Friday's discussion regarding the
'validation' of R, could I ask the list group's opinion on possible
advantages of R over Splus from a pharma/devices perspective? I wish to
exclude the obvious price difference, which doesnât seem to carry as much
weight as I
As I read 21 CFR 11, the regulation deals more with ensuring the security
of the electronic health record itself. Thus, it seemed to me that so long
as the software (SAS, R, Splus, etc.) could not alter the data base in any
way then you're fine (this may be naive, but that's how I understood it).
Not to mention all the work that goes into PROC TEMPLATE and ANNOTATE to
make SAS graphs presentable! I suspect that a lot of companies don't use
SAS graphs or tables at all - they just export the data from SAS to Excel.
-Cody
Cody Hamilton, PhD
Edwards Lifesciences
What I would love to have
The fact that FDA statisticians are using R also assuages one of the main
concerns that I have heard voiced about using R for FDA submissions - that
there would be no statisticians available at FDA to review R code which
would seriously delay the review of a submission.
Mark also brings up a
You can also try validating your regression model via the bootstrap (the
validate() function in the Design library is very helpful). To my mind
that would be much more reassuring than normality tests performed on twenty
residuals.
By the way, be careful with the correlation test - it's only
Following up on Frank's thought, why is it that parametric tests are so
much more popular than their non-parametric counterparts? As
non-parametric tests require fewer assumptions, why aren't they the
default? The relative efficiency of the Wilcoxon test as compared to the
t-test is 0.955, and
Benoit,
Try the aggregate() function.
-Cody
Cody Hamilton, PhD
Edwards Lifesciences
Benoit
Chemineau
Enrico,
prop.test is for testing proportions two at a time. If you want to test
for differences between 4 proportions simultaneously (rather than two at a
time), try a logistic regression model (from which you can get confidence
intervals for each of your groups).
Cody Hamilton, PhD
Staff
Franco,
What about calling the BUGS model below from R using BRUGS?
Regards,
-Cody
francogrex
[EMAIL PROTECTED]
I am developing R code to implement the adaptive design approach of Schafer
and Muller (Stats in Med 2001) for a survival endpoint. Would anyone be
interested in collaborating with me on this code? I have fairly completed
code, but would appreciate someone else's input on my work.
Regards,
Paul,
I believe the model you describe below can be fitted in GENMOD and GLIMMIX
in SAS.
Alternatively, as Brian Ripley suggests, you could use MCMC. BUGS has a
nice example of a multinomial logit model in the second example manual.
While this example considers only fixed effects, it's not
Would this function help:
http://www.csm.ornl.gov/~frome/ES/RRMHex/MHanalysis.txt ?
Regards, -Cody
Frank E Harrell
Jr
Dear Franco,
Have you tried using the beta.start option in MCMClogit? (The problem may
be where you are starting your chain.)
Regards,
-Cody
francogrex
Marc,
I believe the problem is with the function you are passing to the statistic
option. According to the documentation for boot, the statistic option
provides:
A function which when applied to data returns a vector
Has anyone proposed using a bootstrap for Pedro's problem?
What about taking a boostrap sample from x, a boostrap sample from y, take
the difference in the medians for these two bootstrap samples, repeat the
process 1,000 times and calculate the 95th percentiles of the 1,000
computed
John,
I believe the format file must have a .sc or .sas7bcat file extension (not
.sas7bdat which is a sas dataset extension). I think that's why you're
getting the error F:/sas/formats.sc? or formats.sas7bcat not found.
Also, is the carriage return after 'C:/Program in the code below caused
John,
Do you not have sas on your machine? (That would definitely keep sas.get
from executing.)
Regards,
-Cody
John Kane
[EMAIL
John,
According to the sas.get documentation, the format library option must
specify the directory containing the file formats.sct (I don't think it can
be a sas dataset). Also, do you need a forward slash after C: in the
sasprog option below? Finally, sas may not like the period in the file
Thanks for the tip. I will look forward to trying this package out soon!
Regards, -Cody
Hans-Peter
[EMAIL PROTECTED]
Rich Ulrich has compiled some posts (I believe from the S list) on some of
the dangers of stepwise regression:
http://www.pitt.edu/~wpilib/statfaq/regrfaq.html
Regards,
-Cody
Sergio Della
Thank you to all those that responded to Delphine's original post on R and
clinical studies. They have provided much food for thought.
I had a couple of follow up questions/comments. Andrew is very correct in
pointing out that there are classes and workshops available for R. It's my
Thank you to all those that responded to Delphine's original post on R and
clinical studies. They have provided much food for thought.
I had a couple of follow up questions/comments. Andrew is very correct in
pointing out that there are classes and workshops available for R. It's my
Thank you to all those that responded to Delphine's original post on R and
clinical studies. They have provided much food for thought.
I had a couple of follow up questions/comments. Andrew is very correct in
pointing out that there are classes and workshops available for R. It's my
I agree that most problems arise in the data management / file derivation
phase. From my reading of 21 CFR 11, it appears that this document focuses
primarily on data management (as well as on software directly involved in a
medical device) rather than on validation of statistical functions. I
Gregg,
What about
A-data.frame(1,1,1); names(A)=letters[1:3] ; B-matrix(0,2,3)
B-as.data.frame(B)
names(B)-names(A)
rbind(A,B)
-Cody
Gregg Lind
Mat,
Thank you for the update and for the link. I look forward to (hopefully)
using R for future FDA submissions.
Regards,
-Cody
Soukup, Mat
Mat,
Thank you for the update and for the link. I look forward to (hopefully)
using R for future FDA submissions.
Regards,
-Cody
Soukup, Mat
Frederic,
You're performing 8*7/2 = 28 multiple comparisons controlling the FWE at
the .05 level. Using a Bonferroni's adjustment (admittedly more
conservative than the Holm's or Tukey's approach), that's testing each
comparison at the .05/28 = 0.0018 level. With only 100 observations spread
I would like to use R for submissions to FDA/CDRH (the medical device
company I work for currently uses only SAS). Previous postings to the list
regarding R and 21 CFR 11 compliance have been very helpful. However,
reluctance to using open source software for statistical analyses and
reporting
33 matches
Mail list logo