R/symkl.compreg.R

In Compositional: Compositional Data Analysis

################################
#### Symmetric Kullback-Leibler divergence based regression for compositional data
#### Tsagris Michail 5/2015
#### mtsagris@yahoo.gr
#### References: Michail Tsagris (2015)
#### A novel, divergence based, regression for compositional data
#### Proceedings of the 28th Panhellenic Statistics Conference
################################
symkl.compreg <- function(y, x, con = TRUE, B = 1, ncores = 1, xnew = NULL) {
  ## y is dependent variable, the compositional data
  ## x is the independent variable(s)
  ## B is the number of bootstrap samples used to obtain
  ## standard errors for the betas
  ## if B==1 no bootstrap is performed and no standard errors are reported
  ## if ncores=1, then 1 processor is used, otherwise
  ## more are used (parallel computing)
  symkl <- function(para, y, x, d){
    be <- matrix(para, byrow = TRUE, ncol = d)
    mu1 <- cbind( 1, exp(x %*% be) )
    mu <- mu1 / rowSums(mu1)
    sum( (y - mu) * log(y / mu), na.rm = TRUE )
  }
  n <- dim(y)[1]  ## sample size
  x <- model.matrix(y ~ ., data.frame(x) )
  if ( !con )  x <- x[, -1, drop = FALSE]
  p <- dim(x)[2]
  d <- dim(y)[2] - 1  ## dimensionality of the simplex
  namx <- colnames(x)
  namy <- colnames(y)
  if ( is.null( namy ) )  {
    namy <- paste("Y", 2:(d + 1), sep = "")
  } else namy <- namy[-1]
  ## the next lines minimize the symkl function and obtain the estimated betas
  runtime <- proc.time()
  ini <- rnorm( d * dim(x)[2] )
  suppressWarnings({
    qa <- nlm(symkl, ini, y = y, x = x, d = d)
    qa <- nlm(symkl, qa$estimate, y = y, x = x, d = d)
    qa <- nlm(symkl, qa$estimate, y = y, x = x, d = d)
  })
  be <- matrix(qa$estimate, byrow = TRUE, ncol = d)
  covb <- NULL
  runtime <- proc.time() - runtime
  if (B > 1) {
  betaboot <- matrix( nrow = B, ncol = length(ini) )
  nc <- ncores
    if (nc <= 1) {
      runtime <- proc.time()
      for (i in 1:B) {
        ini <- rnorm( d * dim(x)[2] )
        ida <- Rfast2::Sample.int(n, n, replace = TRUE)
        yb <- y[ida, ]
        xb <- x[ida, ]
        qa <- nlm(symkl, ini, y = yb, x = xb, d = d)
        qa <- nlm(symkl, qa$estimate, y = yb, x = xb, d = d)
        qa <- nlm(symkl, qa$estimate, y = yb, x = xb, d = d)
        betaboot[i, ] <- qa$estimate
      }
      covb <- cov(betaboot)
      runtime <- proc.time() - runtime
    } else {
      runtime <- proc.time()
      
      cl <- parallel::makeCluster(ncores)
      # Load required packages on workers
      parallel::clusterEvalQ(cl, library(Rfast2))
      # Export only what workers need
      parallel::clusterExport(cl, 
                             varlist = c("symkl", "y", "x", "n", "d"), 
                             envir = environment())
      
      betaboot <- t( parallel::parSapply(cl, 1:B, function(i) {
        ida <- Rfast2::Sample.int(n, n, replace = TRUE)
        yb <- y[ida, ]  ;  xb <- x[ida, ]
        ini <- rnorm( d * dim(x)[2] )
        suppressWarnings({
          qa <- nlm(symkl, ini, y = yb, x = xb, d = d)
          qa <- nlm(symkl, qa$estimate, y = yb, x = xb, d = d)
          qa <- nlm(symkl, qa$estimate, y = yb, x = xb, d = d)
          qa$estimate  
        })
      })) 
      
      parallel::stopCluster(cl)
      covb <- cov(betaboot)
      runtime <- proc.time() - runtime
    }  ##  end if (nc < 1)
  nam <- NULL
  for (i in 1:p)  nam <- c(nam, paste(namy, ":", namx[i], sep = "") )
  colnames(covb) <- rownames(covb) <- nam
  }  ##  end if (B > 1) {
  if ( !is.null(xnew) ) {
    xnew <- model.matrix(~., data.frame(xnew) )
    if ( !con )  xnew <- xnew[, -1, drop = FALSE]
    mu <- cbind( 1, exp(xnew %*% be) )
    est <- mu / Rfast::rowsums(mu)
    colnames(est) <- colnames(y)
  }  else  est <- NULL
  colnames(be) <- namy
  rownames(be) <- namx
  list(runtime = runtime, be = be, covbe = covb, est = est)
}