R/deadbandBD.R

In deadband: Statistical Deadband Algorithms Comparison

#' deadbandBD Function
#'
#' This function allows you to compute the Bollinger Deadband(BD) algorithm
#' @param x The vector of the samples before the deadband algorithm
#' @param d Deadband percent parameter in range 0..1
#' @param offset How many sample do you want skip at begin? Defaults is n=20
#' @param k multiplier used in Bollinger theory
#' @return A list containing the L2 distance and the Number of filtered samples
#' @keywords BD
#' @export
#' @examples
#' deadbandBD(rnorm(40, mean = 0, sd = 1),0.01,20,2)
deadbandBD <- function(x, d, offset, k){
  bbands <- TTR::BBands( x, n = offset, sd = k)
  Delta <- abs((bbands[,"up"]-bbands[,"dn"])*d)
  x_cache<-x[offset+1]
  NumberOfFiltered = 0
  x_out<-numeric(length(x))
  for(i in (offset+1):(length(x)))
  {
    if( abs(x[i]-x_cache)>Delta[i] )
    {
      x_out[i]=x[i]
      x_cache=x[i]
    }
    else
    {
      NumberOfFiltered=NumberOfFiltered+1
      x_out[i]=x_cache
    }
  }
  # EucDist -> Fidelity
  EucDist = dist(rbind(x, x_out), method = "euclidean")
  # %NumberOfFiltered -> Effectiveness
  NumberOfFiltered = NumberOfFiltered/length(x)
  return(list("NumOfFiltered" = NumberOfFiltered,"EuclideanDist" = EucDist[1]))
}