| 1 | #----------------------------------------------------------------------- |
| 2 | #' Initialize the parameters in a basic way (zero for the conditional mean, |
| 3 | #' uniform for weights, identity for covariance matrices, and uniformly distributed forthe clustering) |
| 4 | #' @param n sample size |
| 5 | #' @param p number of covariates |
| 6 | #' @param m size of the response |
| 7 | #' @param k number of clusters |
| 8 | #' @return list with phiInit, rhoInit,piInit,gamInit |
| 9 | #' @export |
| 10 | #----------------------------------------------------------------------- |
| 11 | basic_Init_Parameters = function(n,p,m,k) |
| 12 | { |
| 13 | phiInit = array(0, dim=c(p,m,k)) |
| 14 | |
| 15 | piInit = (1./k)*rep.int(1,k) |
| 16 | |
| 17 | rhoInit = array(0, dim=c(m,m,k)) |
| 18 | for(i in 1:k) |
| 19 | rhoInit[,,i] = diag(m) |
| 20 | |
| 21 | gamInit = 0.1*array(1, dim=c(n,k)) |
| 22 | R = sample(1:k,n, replace=TRUE) |
| 23 | for(i in 1:n) |
| 24 | gamInit[i,R[i]] = 0.9 |
| 25 | gamInit = gamInit/sum(gamInit[1,]) |
| 26 | |
| 27 | return (data = list(phiInit = phiInit, rhoInit = rhoInit, piInit = piInit, gamInit = gamInit)) |
| 28 | } |