R/utils-stats.R

In cvasi: Calibration, Validation, and Simulation of TKTD Models

# Error functions
#
# Functions measuring the accuracy between original and predicted data. Can
# be used as target functions for e.g. optimization routines which minimize the
# error term.
#
# The following error functions are available:
# * Mean Absolute Error (MAE)
# * Normalized Mean Absolute Error (NMAE)
# * Root Mean Square Error (RMSE)
# * Normalized Root Mean Square Error (NRMSE)
# * Symmetric Mean Absolute Percentage Error (SMAPE)
# * Sum of Squared Errors (SSE)
#
# @rdname errfun
# @param orig numeric vector of original data points
# @param pred numeric vector of predicted data points
# @return numeric error term
# @examples
# # Mean Absolute Error
# mae(1:5, 1:5+0.1)
# # Sum of Squared Errors
# sse(1:5, 1:5+0.1)

# Mean Absolute Error
mae <- function(orig,pred) {
  sum(abs(orig - pred)) / length(orig)
}

# Normalized Mean Absolute Error
# @rdname errfun
# @export
nmae <- function(orig, pred) {
  m <- mean(orig, na.rm=TRUE)
  mae(orig, pred)/ifelse(m == 0, 1, m)
}

# Root Mean Square Error
# @rdname errfun
# @export
rmse <- function(orig, pred) {
  sqrt(sum((orig - pred)^2) / length(orig))
}

# Normalized Root Mean Square Error
# @rdname errfun
# @export
nrmse <- function(orig, pred) {
  rmse(orig, pred)/ifelse(mean(orig) == 0, 1, mean(orig))
}

# Symmetric Mean Absolute Percentage Error
# @rdname errfun
# @export
smape <- function(orig,pred) {
  100 / length(orig) * sum(2 * abs(pred - orig) / (abs(orig) + abs(pred)))
}

# Sum of squared errors
# @rdname errfun
# @export
sse <- function(orig, pred) {
  sum((orig - pred)^2, na.rm = TRUE)
}