StatisticalMethods-0.0.0.1: Statistics/PerformanceMetrics.hs

{-# LANGUAGE DoAndIfThenElse #-}
{-# LANGUAGE RecordWildCards #-}

-- | Common performance metrics which can be calculated using the confusion
-- matrix.
--
-- Fawcett, ROC Graphs: Notes and Practical Considerations for Researchers,
-- 2004, Kluwer Academic Publishers

module Statistics.PerformanceMetrics where

import Control.Monad.Instances

import Statistics.ConfusionMatrix



-- | sensitivity

sensitivity :: ConfusionMatrix -> WrappedDouble
sensitivity ConfusionMatrix{..} = do
  tp <- tp
  fn <- fn
  if tp>0
  then Right $ tp / (tp+fn)
  else Right 0

-- | specificity

specificity :: ConfusionMatrix -> WrappedDouble
specificity ConfusionMatrix{..} = do
  tp <- tp
  fp <- fp
  if tp>0
  then Right $ tp / (tp+fp)
  else Right 0

-- | positive predictive value

ppv :: ConfusionMatrix -> WrappedDouble
ppv ConfusionMatrix{..} = do
  tp <- tp
  fp <- fp
  if tp>0
  then Right $ tp / (tp+fp)
  else Right 0

-- | mathews correlation coefficient

mcc :: ConfusionMatrix -> WrappedDouble
mcc ConfusionMatrix{..} = do
  tp <- tp
  tn <- tn
  fp <- fp
  fn <- fn
  let d' = (tp+fp) * (tp+fn) * (tn+fp) * (tn+fn)
  let d  = if d'==0 then 1 else sqrt d'
  return $ (tp*tn - fp*fn) / d

-- | F-measure

fmeasure :: ConfusionMatrix -> WrappedDouble
fmeasure ConfusionMatrix{..} = do
  tp <- tp
  fp <- fp
  fn <- fn
  if tp>0
  then Right $ 2*tp / (2*tp + fp + fn)
  else Right 0