Gibbs_mix: Gibbs_mix

In echo33-1211/StatComp21093: EM_mix and Gibbs_mix

Description

Gibbs sampling for two-component Gaussian Mixture

Usage

Gibbs_mix(mu1, mu2, sigma1, sigma2, p, N, rv)

Arguments

mu1	the mean of Y1
mu2	the mean of Y2
sigma1	the EM-estimated variance of Y1
sigma2	the EM-estimated variance of Y2
p	the probability of the mixture distribution
N	the number of samples
rv	the random number from mixed normal distribution

Value

the maximum likelihood estimation of the mean, variance and the mixed probablity

Examples

## Not run: 
rv_mix<-function(N,p){
u1<-rnorm(N,0,1)
u2<-rnorm(N,2,1)
j<-runif(N)
k<-as.integer(j>p)
u<-k*u1+(1-k)u2
return (u)
}

p<-0.6
N<-1000
set.seed(2335)
rv<-rv_mix(N,p)
p_0<-0.5
mu1<-0.5
mu2<-1.5
sigma1<-sigma2<-sd(rv)
EM<-EM_mix(mu1,sigma1,mu2,sigma2,p_0,N,rv)
sigma1<-EM[2]
sigma2<-EM[4]
p_hat<-EM[5]
Gibbs_mix(mu1,mu2,sigma1,sigma2,p_hat,N,rv)

## End(Not run)

echo33-1211/StatComp21093 documentation built on Dec. 23, 2021, 11:15 p.m.