cosmetics
[valse.git] / test / helper.R
1 #' Generate a sample of (X,Y) of size n with default values
2 #'
3 #' @param n sample size
4 #' @param p number of covariates
5 #' @param m size of the response
6 #' @param k number of clusters
7 #'
8 #' @return list with X and Y
9 #'
10 generateXYdefault = function(n, p, m, k)
11 {
12 meanX = rep(0, p)
13 covX = diag(p)
14 covY = array(dim=c(m,m,k))
15 for(r in 1:k)
16 covY[,,r] = diag(m)
17 π = rep(1./k,k)
18 #initialize beta to a random number of non-zero random value
19 β = array(0, dim=c(p,m,k))
20 for (j in 1:p)
21 {
22 nonZeroCount = sample(1:m, 1)
23 β[j,1:nonZeroCount,] = matrix(runif(nonZeroCount*k), ncol=k)
24 }
25
26 sample_IO = generateXY(n, π, meanX, β, covX, covY)
27 return (list(X=sample_IO$X,Y=sample_IO$Y))
28 }
29
30 #' Initialize the parameters in a basic way (zero for the conditional mean, uniform for
31 #' weights, identity for covariance matrices, and uniformly distributed for the
32 #' clustering)
33 #'
34 #' @param n sample size
35 #' @param p number of covariates
36 #' @param m size of the response
37 #' @param k number of clusters
38 #'
39 #' @return list with phiInit, rhoInit,piInit,gamInit
40 #'
41 basicInitParameters = function(n,p,m,k)
42 {
43 phiInit = array(0, dim=c(p,m,k))
44
45 piInit = (1./k)*rep(1,k)
46
47 rhoInit = array(dim=c(m,m,k))
48 for (i in 1:k)
49 rhoInit[,,i] = diag(m)
50
51 gamInit = 0.1 * matrix(1, nrow=n, ncol=k)
52 R = sample(1:k, n, replace=TRUE)
53 for (i in 1:n)
54 gamInit[i,R[i]] = 0.9
55 gamInit = gamInit/sum(gamInit[1,])
56
57 return (list("phiInit"=phiInit, "rhoInit"=rhoInit, "piInit"=piInit, "gamInit"=gamInit))
58 }