I applied this code and I get this error over and over again.
library(mixtools)
simulate <- function(lambda=0.3, mu=c(0, 4), sd=c(1, 1), n.obs=10^5) {
x1 <- rnorm(n.obs, mu[1], sd[1])
x2 <- rnorm(n.obs, mu[2], sd[2])
return(ifelse(runif(n.obs) < lambda, x1, x2))
}
x <- simulate()
model <- normalmixEM(x=x, k=2)
Error: object 'C_normpost' not found
Can anyone help me with this?
If on the other hand someone can think of a way to separate two binomial distributions it would be great.
If you have the mixtools package properly installed there should not be an error. I have version 1.1.0 installed which CRAN tells me is the most recent released version. A .C() call to C_normpost is made in three different places in the body of normalmixEM. After that code I see:
str(model) # don't type just `model` ... appears to have no print method for it.
List of 9
$ x : num [1:100000] 3.33 3.17 -2.04 3.66 5.11 ...
$ lambda : num [1:2] 0.299 0.701
$ mu : num [1:2] -0.00253 3.9986
$ sigma : num [1:2] 0.994 1.001
$ loglik : num -197426
$ posterior : num [1:100000, 1:2] 1.94e-03 3.66e-03 1.00 5.17e-04 1.42e-06 ...
..- attr(*, "dimnames")=List of 2
.. ..$ : NULL
.. ..$ : chr [1:2] "comp.1" "comp.2"
$ all.loglik: num [1:39] -427524 -216959 -208812 -204523 -200531 ...
$ restarts : num 0
$ ft : chr "normalmixEM"
- attr(*, "class")= chr "mixEM"
summary(model)
summary of normalmixEM object:
comp 1 comp 2
lambda 0.2992979 0.700702
mu -0.0025302 3.998599
sigma 0.9935503 1.001362
loglik at estimate: -197426.2
So it has done a rather good job of estimating the means (0 and 4) as well as a very good job of estimating the variances of those components (1 and 1). You should probably try to reinstall mixtools in a fresh session and rerun the code.