mer.R

## mre
# Calculates the mean absolute relative error
#mre('predict_Oncorhynchus_mykiss', pars, Data, tW)
##
mre=function(func, p, varargin) {
  #  created: 2010/05/08 by Bas Kooijman, modified 2011/05/02
  
  ## Syntax
  # [merr rerr] = mre(func, p, varargin)
  
  ## Description
  # Calculates the mean absolute relative error, used in add_my_pet
  #
  # Input
  #
  # * func: character string with name of user-defined function;
  #    see <nrregr.html *nrregr*>
    # * p: (np,nc) matrix with p(:,1) parameter values
  # * xywi: (ni,3) matrix with
  #
  #     xywi(:,1) independent variable
  #     xywi(:,2) dependent variable
  #     xywi(:,3) weight coefficients (optional)
  #     The number of data matrices xyw1, xyw2, ... is optional
  #     The first data matrix is assumed to be zero-variate, 
  #       the others uni-variate, which are first reduced to zero-variate data
  #       if all weight coefficients in a uni-variate data-set are zero,
  #       that relative error gets weight zero
  #
  # Output
  #
  # * merr: scalar with mean absolute relative error
  # * rerr: vector with absolute relative errors
  
  i = 1
   nxyw = length(varargin) - 2                                                # number of data sets
   while (i <= nxyw){                                               # loop across data sets
    ci = as.character(i)                                           # character string with value of i
    eval(parse(text=paste('xyw',' = varargin[[i]]', sep='')))                           # assign unnamed arguments to xywi
    if (ncol(xyw) < 3 ){                              # append weights if absent
    xyw = cbind(xyw, rep(1,nrow(xyw))
    }
    if (i == 1) {
      listxyw = paste(xyw,i, sep='') # initiate list xyw
      listf = paste(f, ci, sep='') # initiate list f
    }
    else  {   
      listxyw = c(listxyw,paste(xyw,i, sep='')) # append list xyw
      listf = c(listf,paste(f, ci, sep='')) # append list f
    }
    i = i + 1
  }
  
  # get function values
  listf = func(p[,1], listxyw)
  # abs relative error for zero-variate data
  rerr = cbind(abs(f1 - xyw1[,2]) / max(1e-3, xyw1[,2]), xyw1[,3])
  if (nxyw > 1){ # append uni-variate data
    for (i in c(2:nxyw)) {
      ci = i # character string with value of i
      # rerr = [rerr; sum(abs(fi - xywi(:,2)) .* xywi(:,3) ./ max(1e-3, xywi(:,2)), 1)/ sum(xywi(:,3)) 1];
      eval(['rerr = [rerr; sum(abs(f', ci, ' - xyw', ci, ...
            '(:,2)) .* xyw', ci, '(:,3) ./ max(1e-3, abs(xyw', ci, ...
            '(:,2))), 1)/ sum(max(1e-6, xyw', ci, '(:,3))) 1];']);
      eval(['rerr(end,2) = sum(xyw', ci, '(:,3)) ~= 0;']) # weight 0 if all 0 == xywi(:,3)
    }
  }
  merr = sum(prod(rerr,2))/ sum(rerr(:,2));

}