R/find_T0_multi.R

In SK4FGA: Scott-Knott for Forensic Glass Analysis

#' Calculate T0 for a given list of glass fragments and features.
#'
#' Calculates the "T0" value (the split corresponding to the maximum value of T^2) for a given list of data sets
#' corresponding to glass fragment features assuming they're appropriate values corresponding to glass fragment features.
#'
#' @param data list of glass fragment chemical (or otherwise) features.
#' @param i Starting element (default = 1)
#' @param j Ending element (default = length(array))
#'
#' @return A numeric corresponding to the maximum between-sum-of-squares estimate from the sample.
#' @export find_T0
#'
find_T0 <- function(data, i = 1, j = length(data)){
  # Initialise a vector of T's
  k = j-i+1
  test.T = numeric(k-1)

  for (index in 1:(k-1)){
    g1 = data[1:index]
    g2 = data[(index+1):k]

    d1 = do.call(rbind, g1)
    d2 = do.call(rbind, g2)

    test.T[index] = find_T2(d1, d2)
  }

  T_0 = max(test.T)
  T_0.i = which.max(test.T)

  return(list(x = T_0, i = T_0.i))
}