separate mixture of gamma and normal (with mixtools or ??)
Fitting multicomponent mixtures distributions -- and 3 is already a lot of components -- is inherently ill-conditioned. You may need to reassess your strategy. You might wish to post on stackexchange instead to discuss such statistical issues. Cheers, Bert Bert Gunter "The trouble with having an open mind is that people keep coming along and sticking things into it." -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
On Mon, Jan 23, 2017 at 2:32 PM, Thomas Petzoldt <thpe at simecol.de> wrote:
Dear friends, I am trying to separate bi- (and sometimes tri-) modal univariate mixtures of biological data, where the first component is left bounded (e.g. exponential or gamma) and the other(s) approximately Gaussian. After checking several packages, I'm not really clear what to do. Here is an example with "mixtools" that already works quite good, however, the left component is not Gaussian (and not symmetric). Any idea about a more adequate function or package for this problem? Thanks a lot! Thomas library(mixtools) set.seed(123) lambda <- c(0.25, 0.75) N <- 200 ## dist1 ~ gamma (or exponential as a special case) #dist1 <- rexp(lambda[1]*N, 1) dist1 <- rgamma(lambda[1]*N, 1, 1) ## dist2 ~ normal dist2 <- rnorm(lambda[2]*N, 12, 2) ## mixture x <- c(dist1, dist2) mix <- spEMsymloc(x, mu0=2, eps=1e-3, verbose=TRUE) plot(mix, xlim=c(0, 25)) summary(mix) -- Thomas Petzoldt TU Dresden, Institute of Hydrobiology http://www.tu-dresden.de/Members/thomas.petzoldt
______________________________________________ R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.