module Statistics.Distribution.Hypergeometric
(
HypergeometricDistribution
, hypergeometric
, hdM
, hdL
, hdK
) where
import Data.Binary (Binary)
import Data.Data (Data, Typeable)
import GHC.Generics (Generic)
import Numeric.MathFunctions.Constants (m_epsilon)
import Numeric.SpecFunctions (choose)
import qualified Statistics.Distribution as D
data HypergeometricDistribution = HD {
hdM :: !Int
, hdL :: !Int
, hdK :: !Int
} deriving (Eq, Read, Show, Typeable, Data, Generic)
instance Binary HypergeometricDistribution
instance D.Distribution HypergeometricDistribution where
cumulative = cumulative
instance D.DiscreteDistr HypergeometricDistribution where
probability = probability
instance D.Mean HypergeometricDistribution where
mean = mean
instance D.Variance HypergeometricDistribution where
variance = variance
instance D.MaybeMean HypergeometricDistribution where
maybeMean = Just . D.mean
instance D.MaybeVariance HypergeometricDistribution where
maybeStdDev = Just . D.stdDev
maybeVariance = Just . D.variance
instance D.Entropy HypergeometricDistribution where
entropy = directEntropy
instance D.MaybeEntropy HypergeometricDistribution where
maybeEntropy = Just . D.entropy
variance :: HypergeometricDistribution -> Double
variance (HD m l k) = (k' * ml) * (1 ml) * (l' k') / (l' 1)
where m' = fromIntegral m
l' = fromIntegral l
k' = fromIntegral k
ml = m' / l'
mean :: HypergeometricDistribution -> Double
mean (HD m l k) = fromIntegral k * fromIntegral m / fromIntegral l
directEntropy :: HypergeometricDistribution -> Double
directEntropy d@(HD m _ _) =
negate . sum $
takeWhile (< negate m_epsilon) $
dropWhile (not . (< negate m_epsilon)) $
[ let x = probability d n in x * log x | n <- [0..m]]
hypergeometric :: Int
-> Int
-> Int
-> HypergeometricDistribution
hypergeometric m l k
| not (l > 0) = error $ msg ++ "l must be positive"
| not (m >= 0 && m <= l) = error $ msg ++ "m must lie in [0,l] range"
| not (k > 0 && k <= l) = error $ msg ++ "k must lie in (0,l] range"
| otherwise = HD m l k
where
msg = "Statistics.Distribution.Hypergeometric.hypergeometric: "
probability :: HypergeometricDistribution -> Int -> Double
probability (HD mi li ki) n
| n < max 0 (mi+kili) || n > min mi ki = 0
| otherwise =
choose mi n * choose (li mi) (ki n) / choose li ki
cumulative :: HypergeometricDistribution -> Double -> Double
cumulative d@(HD mi li ki) x
| isNaN x = error "Statistics.Distribution.Hypergeometric.cumulative: NaN argument"
| isInfinite x = if x > 0 then 1 else 0
| n < minN = 0
| n >= maxN = 1
| otherwise = D.sumProbabilities d minN n
where
n = floor x
minN = max 0 (mi+kili)
maxN = min mi ki