I've been at this for a while but am unable to do this efficiently. I want to run many lmer models across multiple variables and extract coefficients from each model to put into a dataframe.
So far I have the following
ivlist<-c("hp", "drat", "wt", "qsec", "vs", "am", "gear", "carb")
mtcars<-mtcars
mtcars$id<- sample(c(1:16), 32, replace=TRUE, prob = c(1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16, 1/16))
mod <- lapply(ivlist, function(x) {
lmer(substitute(mpg ~ cyl + disp + i*disp + (1|id), list(i = as.name(x))), data = mtcars, na.action=na.exclude)
})
res <- lapply(seq_along(mod[c(1:8)]), function(j) {
model = names(mod[[j]])
Vcov <- vcov(mod[[j]], useScale = FALSE)
betas <- fixef(mod[[j]])
se <- sqrt(diag(Vcov))
zval <- betas / se
pval <- 2 * pnorm(abs(zval), lower.tail = FALSE)})
do.call(cbind(names,betas, se, zval, pval), res)
I am now stuck. It's telling me there is no se object. Can you guys help. Even better if you can suggest how i can do this for only the interaction effect.
Thank you in advance
In R
, we get the last output as return
. Here, we may need to return all the objects in a list
or as a named vector
res <- lapply(seq_along(mod[c(1:8)]), function(j) {
model <- names(mod[[j]])
Vcov <- vcov(mod[[j]], useScale = FALSE)
betas <- fixef(mod[[j]])
se <- sqrt(diag(Vcov))
zval <- betas / se
pval <- 2 * pnorm(abs(zval), lower.tail = FALSE)
list(betas = betas, se = se, zval = zval, pval = pval)
})
do.call(rbind, lapply(res, as.data.frame))
# betas se zval pval
#(Intercept) 39.5138536412 2.988611e+00 13.2214778 6.595523e-40
#cyl 0.3492185844 9.057326e-01 0.3855648 6.998190e-01
#disp -0.0805987883 2.376294e-02 -3.3917859 6.943868e-04
#hp -0.1074943073 3.537576e-02 -3.0386435 2.376459e-03
#disp:hp 0.0003166872 1.121111e-04 2.8247615 4.731583e-03
#(Intercept)1 10.6595603973 1.187663e+01 0.8975240 3.694394e-01
#cyl1 -0.8828858263 7.606152e-01 -1.1607522 2.457427e-01
# ...
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.