Thanks for the tips. I'll give them a try!
On Sat, Jun 18, 2011 at 12:07 PM, Dennis Murphy djmu...@gmail.com wrote:
Much better..nice!
Dennis
On Sat, Jun 18, 2011 at 1:53 AM, Dimitris Rizopoulos
d.rizopou...@erasmusmc.nl wrote:
maybe another way is by reconstructing the formula using
Hi,
I would like to do a regression like:
reg - lm(y~log(.), data)
where the log function is applied to . in the form:
log(x1)+ log(x2)+ log(x3)...
instead of in the form
log(x1+x2+x3+...)
Is this possible?
Thank you,
Scott
[[alternative HTML version
Yes, it's possible, but if you want to do prediction on future
x-values, you will likely have a problem.
One way to do it would be something like (assuming y is the first column of dat)
reg - lm(y ~ log(as.matrix(dat[, -1])), dat)
but the output would be pretty ugly (see summary(reg)). Another
maybe another way is by reconstructing the formula using paste(), e.g.,
data - data.frame(y = rnorm(5), x1 = runif(5),
z = runif(5), age = runif(5))
nameRsp - y
nams - names(data)
namsX - nams[!nams %in% nameRsp]
form - as.formula(paste(nameRsp, ~ ,
paste(log(, namsX, ), sep = ,
Much better..nice!
Dennis
On Sat, Jun 18, 2011 at 1:53 AM, Dimitris Rizopoulos
d.rizopou...@erasmusmc.nl wrote:
maybe another way is by reconstructing the formula using paste(), e.g.,
data - data.frame(y = rnorm(5), x1 = runif(5),
z = runif(5), age = runif(5))
nameRsp - y
nams -
5 matches
Mail list logo