bayesian-learning-with-R/E2.dev-040318c.R at master · NCSU-STATS/bayesian-learning-with-R · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
rm(list = ls())
setwd("d:/brucebcampbell-git/bayesian-learning-with-R")

load("E2.RData")
library(R2OpenBUGS)
library(rjags)
library(coda)
library(modeest)

dim(Y1)

cor.y1 <- cor(Y1, use = "pairwise.complete.obs")
cov.y1  <- cov(Y1, use = "pairwise.complete.obs")

N <- nrow(Y1)
p = 6

library("Matrix")
sig <- nearPD(cov.y1)

N = 365
Sigma = sig$mat
Phi = .6
A = matrix(1:p, p, 1)
y = matrix(NA, N,p)
y[1,] = A
for(t in 2:T)
{
  y[t,] = mvrnorm(1, A + Phi * y[t-1,], Sigma)
}

#x <- scale(Y1)

x <- scale(y)


stacks_dat <- list(x=x,p = 6,   N = 365)
mlr_inits <- function() {
  list( rho = 0.00)
}

mlr_model2 <- function(){
  for(i in 1:N)
  {
    theta[i,1:p] ~ dmnorm(x[i,1:p] ,precision2[,])
  }

  # Prior for likelihood parameters: mu2, precision2, rho
  rho  ~  dunif(-1,1)

  for(j in 1:p)
  {
    mu2[j]  ~  dnorm(0,0.01)
  }

  precision2[1:p,1:p] ~ dwish(R[,],k)
  Sigma[1:p,1:p] <- inverse(precision2[,])

  # Missing data model for x
  for(i in 1:N){
    x[i,1:p]~dmnorm(x_mn[],x_prec[,])
  }

  # Priors for missing-data model parameters
  for(j in 1:p){
    x_mn[j]~dnorm(0,0.01)
  }
  x_prec[1:p,1:p]~dwish(R[,],k)
  x_cov[1:p,1:p]<-inverse(x_prec[,])

  k <- p+0.1
  for(j1 in 1:p)
  {
    for(j2 in 1:p)
    {
      R[j1,j2] <- 0.1*equals(j1,j2)
    }
  }
}
n.chains = 1
nSamples=500
samps <- bugs(data = stacks_dat,
              inits = mlr_inits,
              parameters.to.save = c("Sigma"),
              model.file = mlr_model2,
              codaPkg = TRUE,
              n.chains = n.chains, n.burnin=100, n.iter = nSamples, DIC=F)#, n.thin=10

out.coda <- read.bugs(samps)

plot(out.coda)

if(n.chains > 1)
{
  gelman.srf <-gelman.diag(out.coda)
  count.coeff.gt <- sum(gelman.srf$psrf>1.1)
  count.coeff.gt
  #pander(data.frame(count.coeff.gt=count.coeff.gt ),caption="Number of coefficients with a a gelman scale reduction factor greater than 1.1" )
}

chains.ess <- lapply(out.coda,effectiveSize)

first.chain.ess <- chains.ess[1]
plot(unlist(first.chain.ess), main="Effective Sample Size")

chain <- out.coda[[1]]
posterior.means <- vector(365*6)
posterior.modes <- matrix(365*6)
for( i in 1:(365*6) )
{  colname <- colnames(chain)[i]
  print(colname)
  samples <- chain[,i]

  posterior.means[i] <-mean(samples)

  posterior.modes[i] <-mlv(samples)$M
}

plot(posterior.means, posterior.modes)

plot(out.coda)


plot(samps)


library(coda)
library(modeest)
library(MASS)
library(lattice)


densityplot(out.coda)
xyplot(out.coda)