tests/Clustering/FrequentItemSet/BruteForceSpec.hs

   1 {-# LANGUAGE OverloadedLists #-}
   2
   3 module Clustering.FrequentItemSet.BruteForceSpec where
   4
   5 import Control.Arrow (second)
   6 import Control.Monad (forM_)
   7 import Data.Function (($), (.))
   8 import Data.Functor ((<$>))
   9 import Data.Int (Int)
  10 import Data.List qualified as List
  11 import Data.Ord (Ord)
  12 import Data.Ratio (Rational, (%))
  13 import Data.Semigroup (Semigroup (..))
  14 import GHC.IsList (toList)
  15 import Logic
  16 import Numeric.Probability (assertProbability)
  17 import System.FilePath ((<.>))
  18 import Test.Syd
  19 import Text.Show (Show (..))
  20 import Prelude (Num)
  21
  22 import Clustering.FrequentItemSet.BruteForce
  23 import Utils
  24
  25 -- | From https://research.nii.ac.jp/~uno/code/lcm.html#IntroductionstoFrequentItemsetMining
  26 databaseTakeakiUno :: Ord item => Num item => Transactions item
  27 databaseTakeakiUno =
  28   [ [1, 2, 5, 6, 7]
  29   , [2, 3, 4, 5]
  30   , [1, 2, 7, 8, 9]
  31   , [1, 7, 9]
  32   , [2, 7, 9]
  33   , [2, 7, 9] -- Copy-paste typo on the original example
  34   , [1, 9] -- Add this to increase the support of [1,9] because the original example is wrong…
  35   , [2]
  36   ]
  37
  38 -- | From https://hal.science/hal-03500847
  39 databaseHAL03500847T2 :: Ord item => Num item => Transactions item
  40 databaseHAL03500847T2 =
  41   [ [1, 3, 7, 6]
  42   , [1, 2, 7]
  43   , [2, 8, 9, 10]
  44   , [5, 11]
  45   ]
  46
  47 spec :: Spec
  48 spec = do
  49   describe "allFrequentItemSets" do
  50     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  51       golden ("db=TakeakiUno" <.> "minSupp=" <> show minSupp) $
  52         -- Alas, this is not a zero-cost `coerce`
  53         unName <$> allFrequentItemSets @Int (unitName databaseTakeakiUno) (assertStrictlyPositive minSupp)
  54     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  55       golden ("db=HAL03500847T2" <.> "minSupp=" <> show minSupp) $
  56         unName <$> allFrequentItemSets @Int (unitName databaseHAL03500847T2) (assertStrictlyPositive minSupp)
  57
  58   describe "associationRules" do
  59     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  60       forM_ ([(75 % 100)] :: [Rational]) \minConf ->
  61         golden
  62           ("db=TakeakiUno" <.> "minSupp=" <> show minSupp <.> "minConf=75%")
  63           [ associationRules fis (unitName databaseTakeakiUno) (unitName (assertProbability minConf))
  64           | fis <- allFrequentItemSets @Int (unitName databaseTakeakiUno) (assertStrictlyPositive minSupp)
  65           ]
  66
  67   describe "allClosedFrequentItemSets" do
  68     forM_ ([2 .. 3] :: [Int]) \minSupp ->
  69       forM_ ([2 .. 3] :: [Int]) \minSize ->
  70         golden ("db=TakeakiUno" <.> "minSupp=" <> show minSupp <.> "minSize=" <> show minSize) $
  71           second (List.sort . toList) . unName
  72             <$> allClosedFrequentItemSets @Int (unitName databaseTakeakiUno) (assertStrictlyPositive minSupp) (assertStrictlyPositive minSize)
  73     forM_ ([1 .. 3] :: [Int]) \minSupp ->
  74       forM_ ([2 .. 4] :: [Int]) \minSize ->
  75         golden ("db=HAL03500847T2" <.> "minSupp=" <> show minSupp <.> "minSize=" <> show minSize) $
  76           second (List.sort . toList) . unName
  77             <$> allClosedFrequentItemSets @Int (unitName databaseHAL03500847T2) (assertStrictlyPositive minSupp) (assertStrictlyPositive minSize)