| 1 | #' Generate a sample of (X,Y) of size n |
| 2 | #' @param meanX matrix of group means for covariates (of size p) |
| 3 | #' @param covX covariance for covariates (of size p*p) |
| 4 | #' @param covY covariance for the response vector (of size m*m*K) |
| 5 | #' @param pi proportion for each cluster |
| 6 | #' @param beta regression matrix, of size p*m*k |
| 7 | #' @param n sample size |
| 8 | #' |
| 9 | #' @return list with X and Y |
| 10 | #' @export |
| 11 | generateXY = function(meanX, covX, covY, pi, beta, n) |
| 12 | { |
| 13 | p = dim(covX)[1] |
| 14 | m = dim(covY)[1] |
| 15 | k = dim(covY)[3] |
| 16 | |
| 17 | X = matrix(nrow=n,ncol=p) |
| 18 | Y = matrix(nrow=n,ncol=m) |
| 19 | class = matrix(nrow = n) |
| 20 | |
| 21 | require(MASS) #simulate from a multivariate normal distribution |
| 22 | for (i in 1:n) |
| 23 | { |
| 24 | class[i] = sample(1:k, 1, prob=pi) |
| 25 | X[i,] = mvrnorm(1, meanX, covX) |
| 26 | print(X[i,]) |
| 27 | print(beta[,,class[i]]) |
| 28 | Y[i,] = mvrnorm(1, X[i,] %*% beta[,,class[i]], covY[,,class[i]]) |
| 29 | } |
| 30 | |
| 31 | return (list(X=X,Y=Y, class = class)) |
| 32 | } |
| 33 | |
| 34 | #' Generate a sample of (X,Y) of size n with default values |
| 35 | #' @param n sample size |
| 36 | #' @param p number of covariates |
| 37 | #' @param m size of the response |
| 38 | #' @param k number of clusters |
| 39 | #' @return list with X and Y |
| 40 | #' @export |
| 41 | generateXYdefault = function(n, p, m, k) |
| 42 | { |
| 43 | rangeX = 100 |
| 44 | meanX = rangeX * matrix(1 - 2*runif(p*k), ncol=k) |
| 45 | covX = array(dim=c(p,p,k)) |
| 46 | covY = array(dim=c(m,m,k)) |
| 47 | for(r in 1:k) |
| 48 | { |
| 49 | covX[,,r] = diag(p) |
| 50 | covY[,,r] = diag(m) |
| 51 | } |
| 52 | pi = rep(1./k,k) |
| 53 | #initialize beta to a random number of non-zero random value |
| 54 | beta = array(0, dim=c(p,m,k)) |
| 55 | for (j in 1:p) |
| 56 | { |
| 57 | nonZeroCount = sample(1:m, 1) |
| 58 | beta[j,1:nonZeroCount,] = matrix(runif(nonZeroCount*k), ncol=k) |
| 59 | } |
| 60 | |
| 61 | sample_IO = generateXY(meanX, covX, covY, pi, beta, n) |
| 62 | return (list(X=sample_IO$X,Y=sample_IO$Y)) |
| 63 | } |
| 64 | |
| 65 | #' Initialize the parameters in a basic way (zero for the conditional mean, uniform for weights, |
| 66 | #' identity for covariance matrices, and uniformly distributed for the clustering) |
| 67 | #' @param n sample size |
| 68 | #' @param p number of covariates |
| 69 | #' @param m size of the response |
| 70 | #' @param k number of clusters |
| 71 | #' @return list with phiInit, rhoInit,piInit,gamInit |
| 72 | #' @export |
| 73 | basicInitParameters = function(n,p,m,k) |
| 74 | { |
| 75 | phiInit = array(0, dim=c(p,m,k)) |
| 76 | |
| 77 | piInit = (1./k)*rep(1,k) |
| 78 | |
| 79 | rhoInit = array(dim=c(m,m,k)) |
| 80 | for (i in 1:k) |
| 81 | rhoInit[,,i] = diag(m) |
| 82 | |
| 83 | gamInit = 0.1 * matrix(1, nrow=n, ncol=k) |
| 84 | R = sample(1:k, n, replace=TRUE) |
| 85 | for (i in 1:n) |
| 86 | gamInit[i,R[i]] = 0.9 |
| 87 | gamInit = gamInit/sum(gamInit[1,]) |
| 88 | |
| 89 | return (list("phiInit" = phiInit, "rhoInit" = rhoInit, "piInit" = piInit, "gamInit" = gamInit)) |
| 90 | } |