Calculate true positive rate, false positive rate & false discovery rate from contingency table in R

calc_tpr_fpr_fdr.R

## Suppose we have a contigency table tbl formed by the table(...) command in R, with
## a logical vector of discoveries as the first argument and a logical vector of
## the ground truth as the second, e.g. tbl <- table(discoveries, ground_truth), then
## this function calculates the true positive rate, false positive rate and false discovery rate
## as per the wikipedia definition at https://en.wikipedia.org/wiki/Sensitivity_and_specificity

calculate_statistics <- function(tbl) {
  P <- sum(tbl[,2])
  N <- sum(tbl[,1])
  TP <- tbl[2,2]
  TN <- tbl[1,1]
  FP <- tbl[1,2]
  FN <- tbl[2,1]
  TPR <- TP / P
  FPR <- FP / N
  FDR <- FP / (TP + FP)
  data.frame(P = P, N = N, TP = TP, TN = TN, FP = FP, 
             FN = FN, TPR = TPR, FPR = FPR, FDR = FDR)
}

I think the off-diagonals (false-positive, false-negative) are not correct per the confusion matrix, and should instead be:

  FP <- tbl[2,1]
  FN <- tbl[1,2]